Monitoring and video processing systems and methods

Abstract

The present invention relates to monitoring and / or video processing systems and methods. Particular applications are in the field of outdoor performance tracking and analytics, particularly fishing performance tracking and analytics. In a particular case the present invention relates to a fishing system designed to automatically track fishing events and / or determine fishing performance, boating analytics, or process fishing related video. In one example the present invention relates to a monitoring system for a fishing rod, the monitoring system comprising: one or more sensors configured to monitor an area including at least one fishing rod, and a processor configured to: receive an output from the one or more sensors; determine one or more regions of at least one fishing rod from the output; determine one or more events based on the output in the one or more regions of the at least one fishing rod.

Description

[0001] MONITORING AND VIDEO PROCESSING SYSTEMS AND METHODS TECHNICAL FIELD

[0002] The present invention relates to monitoring and / or video processing systems and methods. Particular applications are in the field of outdoor performance tracking and analytics, particularly fishing performance tracking and analytics. In a particular case the present invention relates to a fishing system designed to automatically track fishing events and / or determine fishing performance, boating analytics, or process fishing related video. BACKGROUND OF THE INVENTION

[0003] In particular examples the present disclosure relates to the field of fishing performance tracking and analytics. Traditional methods of tracking fishing performance, such as manual logging and wearable devices, are effort-intensive and prone to human error. Manual logging requires anglers to record data in notebooks or apps, which is timeconsuming and often inaccurate. Wearable devices, such as watches and rod-mounted sensors, attempt to detect casting motions of a user but require the user to wear a device or modify their fishing gear, disrupting the natural fishing movement.

[0004] It is an object of the present invention to provide a monitoring system that automates or simplifies one or more processes for monitoring performance of an activity, such as fishing.

[0005] It is an object of the present invention to provide a video processing system that automates or simplifies one or more processes for collating videos of performance of an activity, such as fishing.

[0006] It is an object of the present invention to provide a fishing system that provides an automated, non-intrusive solution that collects real-time analytics without requiring active engagement. An additional object of the present invention is to provide a system that uses sensors to automatically detects fishing events, records key events to generate trip reports for real-time and / or post-trip analytics. It is envisaged that the invention, or preferred embodiments thereof, may comprise additional features which further improve its performance and cost-effectiveness. However, this is not intended to be limiting, and the invention may or may not incorporate such additional features.

[0007] Alternatively, it is an object of the present invention to at least provide the public with a useful choice.

[0008] SUMMARY OF THE INVENTION

[0009] P1142NZ00This international application claims priority to and incorporates by reference in its entirety New Zealand Provisional Patent Application No. 820612, filed 15 April 2025. A first disclosure comprises a fishing system for tracking fishing performance over time, comprising:

[0010] a sensor module configured to automatically capture performance data in real-time during on-water experiences;

[0011] a real-time analytics component configured to receive the captured performance data and calculate outcome insights;

[0012] a processor to analyse and store the captured performance data and / or outcome insights in one or more historical record;

[0013] wherein the one or more historical record is configured to generate forecasts or recommendations for use in current and / or future fishing experiences.

[0014] Optionally the captured performance data includes detecting one or more of the following: casting motions, fishing rod activity, fish fighting dynamics, identification and / or classification of fish species, measurement of fish caught, compliance with regulations, identification and / or classification of a lure type, and / or detection of environmental conditions.

[0015] Optionally the sensor module comprises at least one sensor, the sensor may include a mmWave radar sensor; a vision-based sensor, such as a camera, and infrared based sensor, a sound sensor or a combination thereof. More preferably the vision-based sensor uses Al to capture and log visual activities.

[0016] A first disclosure comprises a fishing system for tracking fishing performance, comprising: a mmWave radar sensor configured to detect casting motions, rod activity, and fish fighting dynamics; a vision-based Al camera configured to classify fish species, lure types, and detect environmental conditions in real-time; and a processor configured to process sensor data locally for instant analytics and offline capability.

[0017] Optionally the sensor module is triggered to automatically capture performance data in response to detection of an event. The event may include any one of a strike, hook-up, casting motion, fish landing, rod activity, boat movement, an activating voice command and / or specific gesture.

[0018] Optionally the processor is configured to generate trip records by aggregating environmental conditions, fishing performance metrics, and automatically edited video highlights.A further disclosure comprises a fishing system for tracking fishing performance, the fishing system comprising:

[0019] • a non-contact sensor configured to detect an object's position, size, distance, and / or velocity;

[0020] • a visual sensor configured to classify one or more of a fish species, a lure types, and / or detect environmental conditions in real-time; and

[0021] • a processor configured to process sensor data locally for instant analytics, low latency, and offline capability;

[0022] wherein the fishing system is configured to automatically detect and log key fishing events, generate trip records, and synchronize with external devices.

[0023] Optionally the non-contact sensor is a millimetre-wave (mmWave) radar or other standalone or hybrid sensor.

[0024] Optionally the visual sensor includes a camera that uses artificial intelligence (Al) to classify fish species, fish size, lure types, and / or detect other environmental conditions in real-time.

[0025] Optionally the processor is configured to generate comprehensive trip records by aggregating environmental conditions, fishing performance metrics, and automatically edited video highlights. The processing could be done on device, or devices or in the cloud or a combination of on device and / or in the cloud.

[0026] Optionally the fishing system further includes a sound module, preferably the sound module includes at least one of a speaker and / or microphone configured to send and / or receive verbal communication that are processed by the processor.

[0027] Optionally the system is configured to synchronize seamlessly with smartphones, marine electronics, and cloud-based analytics.

[0028] Optionally the system is configured to detect motion and objects using sensors, including mmWave radar and vision-based Al systems.

[0029] Optionally the system is configured to provide real-time insights and post-trip analysis, ensuring users can focus entirely on fishing without being burdened by data logging.

[0030] Optionally the system is configured to trigger video recording automatically when an event is detected. The event may include any one of a strike, hook-up, casting motion, rod activity, boat movement, an activating voice command and / or specific gesture.Optionally the system is configured to generate trip records that aggregate environmental conditions, fishing performance metrics, and video content. In some forms the processor is configured to publish, send notification updates or livestream one or more of the instant analytics, detected key fishing events, videos and / or trip record.

[0031] Optionally the system is configured to provide Al-powered insights with actionable recommendations based on historical data and real-time conditions.

[0032] A further disclosure comprises a fishing system for a fishing rod, the monitoring system comprising:

[0033] a processor configured to:

[0034] receive an output from the one or more sensors;

[0035] determine one or more regions of at least one fishing rod from the output;

[0036] determine one or more events based on the output in the one or more regions of the at least one fishing rod.

[0037] Optionally, the system comprises the one or more sensors, the one or more sensors configured to monitor an area including at least one fishing rod, and

[0038] Optionally, the determination of an event comprises determining the position and / or movement of the fishing rod. Optionally, the determination of an event comprises determining a position or movement of the fishing rod in the one or more regions.

[0039] Optionally, the movement may comprise one or more of: tip twitch, shaft deflection, and

[0040] tension variations in the line. Tension variations may be between states, such as from slack to taut. Optionally, the movement comprises one or more of: High-frequency motion; Line jumps, twitches or lateral movement; Rotational motion of reel components (handle / spool); Rates of change of curvature; and Persistence of bend patterns.

[0041] Optionally, the position comprises one or more of: Curvature changes along the rod shaft, and Changes in the tension of the fishing line.

[0042] Optionally, the determination of an event is based, at least in part, on the position and / or movement of the fishing rod in each of the one or more regions. Optionally the position and / or movement is over a period of time. Optionally, the period of time comprises the length of a video segment, a time period between one more images, and / or a time period of a series of outputs from the sensor.

[0043] Optionally, the output from the one or more sensors is a video and / or a plurality of images.Optionally, the processor is configured to timestamp the output. Optionally, the output has an associated timestamp. Optionally, the one or more events comprise a time characteristic. The time characteristics may be any one or more of: a time stamp, a start time, and end time and / or a time period associated with the output and / or event. Optionally, the event time characteristic is determined based on one or more time characteristics of the outputs. The time characteristic may be a time stamp. Optionally, the processor is configured to associated additional sensor data with the one or more events based on the time characteristic.

[0044] Optionally, data from one or more additional sensors is used to determine the one or more events. Optionally, additional sensors comprise one or more of image sensors, video sensors, audio sensors; radar, sonar, lidar, movement sensors, and location sensors.

[0045] Optionally, the determination of an event uses the output of at least two of the regions. Optionally, the determination of an event uses the output of at least three, four or five regions. Optionally, the regions comprise one or more of a rod tip zone; a mid-shaft zone; a butt zone; a line zone; and a reel zone. Optionally, the regions comprise an angler zone and / or a fishing rod zone.

[0046] Optionally, the processor is configured to monitor at least one additional item of fishing gear. Optionally, the item comprises an outrigger and / or a downrigger and / or a support frame. Optionally, the item interacts with one or more of the line and fishing rod.

[0047] Optionally, each of the regions comprise a portion of the fishing rod. Optionally, the regions overlap and / or abut so as to substantially cover the fishing rod.

[0048] Optionally, the system comprises one or more models. Optionally, the processor is configured to use the one or more models to determine the one or more regions and / or determine the one or more events. Optionally, the models are trained models.

[0049] Optionally, the trained model(s) is an artificial intelligence model. Optionally, artificial intelligence comprises computer-vision processes, CNN based architectures, multimodal approaches, vision-language models, large language models. Optionally, the artificial intelligence model is configured to receive a plurality of images as input and output structured event descriptions.

[0050] Optionally, the processor is configured to determine the one or more regions of the at least one fishing rod using a first trained model.Optionally, the first model is trained by training data comprising a plurality of images including fishing rods where one or more of the one or more regions have been labelled. Optionally, the regions are labelled as an area of the image.

[0051] Optionally, the images include a plurality of fishing rods. In some case, the images are divided into images each having one fishing rod.

[0052] Optionally, the model is trained to determine a first of the one or more regions and determine at least a second of the one or more regions based on the location of the first region.

[0053] Optionally, the processor is configured to apply: a first model configured to determine the one or more regions of the fishing rod, and a second model configured to determine the one or more events based on the determined the one or more regions of the first model.

[0054] Optionally, the first and / or the second model are trained models.

[0055] Optionally, the one or more regions identified in the first model are input into the second model.

[0056] Optionally, the second model is configured to identify events, based on at least two or more of the regions. Optionally, the second model is configured to identify one or more events based on movements in two or more of the regions concurrently, or within a set timeperiod.

[0057] Optionally, the same training set is used to train the model for region detection and event detection. Optionally, models are trained with the same training set and different reward functions.

[0058] Optionally, the training data is used to train both a first model for determining regions and a second model for determining events. Optionally, the training data set is configured to train multiple models, or a single model completing multiple tasks.

[0059] Optionally, the first model comprises a vision model configured to detect and label the one or more regions of the fishing rod.

[0060] Optionally, the first model comprises a computer vision pipeline or a convolutional neural network.

[0061] Optionally, the first model is trained on training data comprising a plurality of images (such as images or video frames) labelled to identify the one or more regions.

[0062] Optionally, the second model is trained on training data comprising a plurality of images (such as images or video frames) labelled to identify events based on the one or more regions.Optionally, a model is trained to determine the one or more events from the received outputs. Optionally, the model is trained on training data comprising a plurality of images labelled with the one or more regions and the one or more events. Optionally, the model is trained on training data comprising a plurality of images labelled with the one or more events, only the one or more events, or without the one or more regions of the fishing rod.

[0063] Optionally, the processor is configured to determine the one or more events analytically. Optionally, the analytic determination is based on the kinematics of the fishing rod and / or line in each of the one or more regions.

[0064] Optionally, one or more of the models uses machine learning and / or artificial intelligence. Optionally, one or more of the models uses one or more of: computer-vision processes, CNN based architectures, multi-modal approaches, vision-language models, large language models. Optionally, the artificial intelligence model is configured to receive a plurality of images as input and output structured event descriptions.

[0065] Optionally, the processor is configured to: determine one or more occluded regions are not present in the output from the sensor, and infer the movement of the fishing rod in the occluded regions, based on one or more other regions of the fishing rod, and / or determine one or more events of the fishing rod based on one or more other regions of the fishing rod, and / or determine one or more events of the fishing rod based output from the occluded region before and after occlusion.

[0066] Optionally, the training data comprises a plurality of images with at least one of the regions occluded.

[0067] Optionally, the training data comprises a plurality of images where one or more of the regions of the fishing rod are occluded. Optionally, the occlusions are for a least a portion or all of a series of images.

[0068] Optionally, the training data comprises one or more image sequences in which one or more of the regions are occluded for only a portion of the sequence. Optionally, the image sequence comprises at least one image before occlusion. Optionally, the image sequence comprises at least one image after occlusion.

[0069] Optionally, the processor is configured to determine event sequences. Optionally, event sequences comprise a series of events from an initial event to a final event. Optionally, the event sequences typically reflect a fishing action. Optionally, a model is configured to determine event sequences. Optionally, the processor is configured to determine eventsequences algorithmically based on the detected events. Optionally, the processor determines if determined events are initial, intermediate, or final events.

[0070] Optionally, the model is trained to determine individual events and to determine event sequences.

[0071] Optionally, an event sequence comprises an initial event, one or more intermediate events, and a final event. Optionally, the initial event comprises one or more of: a cast, bait deployment, or a strike. Optionally, the intermediate events comprise one of more of: a strike, fight update, a bite. Optionally, the final event comprises one or more of: a landed fish, a lost fish, missed strike, wind-in or a retrieve. Optionally, a final event is determined by an inactivity time-out.

[0072] Optionally, an event sequence is determined by the model. For example, the processor may be trained by labelled training data of event sequences to determine initial and final events and / or event sequences. Optionally, the processor is configured to initialise an event sequence when an initial event is determined. Optionally, the processor is configured to add to the event sequences as intermediate events are detected. Optionally, the processor is configured to end an event sequence when a final event occurs. Optionally, one or more images and / or videos are associated with each event.

[0073] Optionally, the processor is configured to initialise separate event sequences for each fishing rod.

[0074] Optionally, the processor is configured to determine if a determined event is a continuation of a current event sequence. Optionally, the processor is configured to associate the determined event to the current event sequences. Optionally, the determination is based event characteristics. Optionally, event characteristics comprise temporal and / or spatial proximity. Optionally, the processor is configured to determine if the determined event is a final event.

[0075] Optionally, the processor is configured to determine if a determined event is the initial event of a sequence of events. Optionally, the processor is configured to initiate a sequence of events and await determination of continuing events after the initial event. Optionally, the processor is configured to manage multiple active sequences of events currently.

[0076] Optionally, the processor is configured to determine a quality characteristic of an event and / or event sequence. Optionally, the quality characteristic is an importance characteristic and / or an excitement characteristic. The quality characteristic may reflect an expected importance of the event and / or event sequence for later processing. Optionally, theprocessor system tracks and / or updates a quality characteristic of an event sequence as new events are added to the event sequence. Optionally, the processor processes the event based, in part, on the quality characteristic. Optionally, the processor determines an amount of the output (e.g. images and / or video) to save based on the quality characteristic.

[0077] Optionally, the quality characteristic is a numerical score. Optionally, events, or sensor data increase or decrease the numerical score. Optionally, events received from other sensors affect the numerical score. Optionally, the quality characteristic is continuously updated. Optionally, the quality characteristic is averaged over a time window. Optionally, the quality characteristic comprises a weighted value.

[0078] Optionally, the quality characteristic and / or weighted value comprises inputs from the sensor output and / or one or more additional sensors.

[0079] Optionally, additional sensors comprise one or more of: angler position sensors, audio sensors, reel noise sensors, water sounds.

[0080] Optionally, the processor is configured to process event data based on the quality characteristic. Optionally, the processor is configured to store more event data (for example, video, or sensor output) for events or sequences of events with higher quality characteristics. Optionally, the processor is configured to store more event data for time periods of the events or sequences of events with higher importance.

[0081] Optionally, reducing the event data comprises reducing stored sensor output (e.g. video) size, quality or frame rate. Optionally, reducing the event data comprises retaining only images (instead of, for example, video).

[0082] Optionally, a user can set a profile for actions based on quality characteristic.

[0083] Optionally, a profile sets thresholds for actions based on quality characteristic.

[0084] Optionally, training data is provided from multiple sensor positions. Optionally, sensor positions comprise one of more of: the stern of the boat; the bow of the boat; body worn; mounted to the fishing rod; external to the boat.

[0085] Optionally, sensor positions are selected based on their field of view. Optionally, each sensor comprises a layout region. Optionally, a layout region refers to a repeatable area of the sensor field of view. Optionally, a sensor comprises multiple layout regions. Optionally, a sensor comprises separate layout regions for each fishing rod on a boat, or each active fishing rod on a boat.

[0086] Optionally, the processor is configured to detect one or more fishing rods. Optionally, the processor is configured to: firstly, detect each fishing rod (optionally, or layout regionsfor fishing rods), and secondly identify regions on each detected rod. Optionally, the processor is configured to determine if a fishing rod is active and / or inactive. Optionally, only fishing rods determined as active are processed by the processor.

[0087] Optionally, the sensor comprises one or more of an image sensor, a video sensor or a range sensor. The image sensor or video sensor may comprise a camera. The range sensor may comprise a lidar or radar. For example, a mmWave radar. Optionally, the training data is selected based on sensor type. Optionally, training data for a range sensor comprises labelled range sensor outputs. Optionally, the training data is not sensor specific.

[0088] Optionally, a plurality of sensors are used. In some case a first sensor is used for event detection and a second sensor for analysis of the detected event. Optionally, a single sensor is used for event detection and analysis. Optionally, a video camera is used for event detection, and a range sensor is used to determine characteristics of a fish. Characteristics may include as length, size, weight, type.

[0089] Optionally, training data is provided from each of a plurality of sensor positions. Optionally, the training data is labelled with the sensor position. Optionally, the training data is agnostic to sensor position and / or type.

[0090] Optionally, the one or more sensors are moveable. Optionally, the one or more sensors are configured to track the fishing rod. Optionally, the sensors comprise a motor configured to move the sensor. Optionally, the sensors a mounted on a gimbal. Optionally, the sensors are configured to follow one or more fishing rod. Optionally, following comprises rotating or changing the field of view to maintain a view of the fishing rod. Optionally, the regions available to a sensor and / or the ability to determine events based on the regions are adjusted based on the location and / or field of view of the sensor.

[0091] Optionally, events comprise any one or more of casts, bites, subtle bites, strikes, (light strikes, heavy strikes,) landings, fights (sustained, short), lost fish, and misses.

[0092] Optionally, one or more characteristics of the one or more events are determined. Optionally, the one or more characteristics include: a strength, a time, and a size.

[0093] Optionally, the processor is configured to determine a completion of the one or more events based on the output in the one or more regions of the at least one fishing rod.

[0094] Optionally, training data for the end of an event comprises one or more labelled images showing: short term gear stillness, visible fish on or near the boat

[0095] Optionally, the processor associates one or more fishing rods with one or more users. Optionally, the processor determines the user of a fishing rod from stored data. Optionally,the processor determines the user in proximity to the fishing rod. Optionally, the processor associates the events and / or performance metrics of each fishing rod to the associated user or users.

[0096] Optionally, the processor is configured to analyse a plurality of detected events and determine one or more metrics based on the plurality of events. The metrics may comprise: fishing time, success rate, number of bites, number of fights, casting rate, casting length; number of casts, catch success rates and / or any suitable metric.

[0097] Optionally, the processor is configured to determine one or more metrics for an event and / or a sequence of events. Optionally, the metrics comprise one or more characteristics of the event. Optionally, the metrics comprise one or more measurements from the one or more regions of the fishing rod. Optionally, the metrics comprise one or more characteristics of the sequence of events.

[0098] Optionally, characteristics of an event (or metrics) comprise one or more: per rod position statistics, where the cast landed relative to the boat, approximate distance, the technique and retrieve pattern, retrieve speed, fish per cast ratio, and

[0099] strike to cast ratios.

[0100] Optionally, the system is configured to determine the effect of one or more characteristics of the event on the outcome of the event. Optionally, the system is configured to determine characteristics on event sequences.

[0101] Optionally, the one or more sensors are mounted to a person, a vehicle (such as a boat), and / or the ground. Optionally, the one or more sensors are not mounted on the fishing rod. Optionally, the system is separate to the sensors. Optionally, the one or more sensors are positioned next to a body of water. Optionally, the one or more sensors are positioned on a mounting at the edge of a body of water.

[0102] Optionally the system is a dedicated module connected to a boat or a module configured to be integrated into a boat or a sub-module of a boat system. Optionally, the one or more sensors comprise shore mounted cameras, third party action cameras, angler worn devices, or combinations thereof.

[0103] Optionally, the system comprises a vehicle position sensor. Optionally, the vehicle position sensor comprises inertial measurement unit (IMU). Optionally, the processor is configured to determine vehicle (boat) movement based on the output of the vehicle position sensor and: process the output of the sensor to remove movement of the vehicle, and / or process the output of the sensor to determine movement of the fishing rod independent fromthe movement of the vehicle, and / or determine the relative movement of the fishing rod and the vehicle.

[0104] Optionally, the processor is configured to determine vehicle movement based on the output of the vehicle position sensor and stabilise the output of the sensor.

[0105] Optionally, processor determines the one or more event based, in part, on the output of the sensor for vehicle movement. Optionally, the processor is configured to separate whole-body rotation of the fishing rod (and / or a user of the fishing rod) from additional motion of the fishing rod or line caused by, for example a cast, hook set or fight.

[0106] Optionally, the vehicle position sensor detects movement on at least 3 or at least 6-axes.

[0107] Optionally, the processor is configured to determine high-frequency movement in each of the regions. Optionally, the processor is configured to apply a high pass filter to concentrate on high-frequency movement in each of the regions. Optionally, a high pass filter is used to remove or reduce the effects of vehicle motion. Optionally, the high-frequency movements (or the cutoff frequency for the high pass filer) may be considered as above 1HZ, or above 0.5Hz, above 2Hz, above 5Hz, or above 10Hz.

[0108] Optionally, the processor is configured to associate one or more sensor outputs from additional sensors with the determined event. Optionally, the processor is configured to associate the one or more sensor outputs with the determined event based on time overlap. Optionally, the processor is configured to compare an event time period with a time period or time stamp associate with the one or more sensor outputs. Optionally, the processor is configured to create an event based on the one or more sensor outputs from additional sensors. Optionally, the processor is configured to modify an event based on the one or more sensor outputs from additional sensors.

[0109] Optionally, one or more additional sensors comprise a location sensor (e.g., GPS), an environmental sensor, sensors on fishing accessories, accessory sensors, manual (or user) inputs, cameras, personal electronic device sensors, cameras, image sensor, depth sensors, sensors from other boat systems.

[0110] Optionally, one or more additional sensor outputs are provided by computer-vision models. Optionally, the computer-vision models are configured to determine a characteristic of an event and / or a catch.

[0111] Optionally, the processor is configured to determine equipment on or associated with the fishing rod and store this with the event. Optionally, the processor is configured toassociate one of more environmental conditions with the event. Optionally, the system comprises one or more environmental sensors. Optionally, the processor is configured to determine a movement of a person and / or animal. The person may be a user of the fishing rod. Optionally, the processor is configured to associate the person and / or animal with an event.

[0112] Optionally, the processor is configured to determine one or more characteristics of the animal, optionally including: Fish species, Fish measurement, Fish compliance. Optionally, the processor is configured to receive the one or more characteristics.

[0113] Optionally, the processor is configured to determine one or more event or environment characteristics, optionally including: water conditions, Lure type, cast type, and Environmental conditions.

[0114] Optionally, the system comprises a microphone, wherein the processor is configured to receive one or more audio signals from the microphone. Optionally, the processor is configured to determine one or more events from characteristic signals in the audio signal, optionally in combination with the sensor outputs.

[0115] Optionally, the processor is configured to determine on or more of: a characteristic audio signal of the rod, a characteristic audio signal of the water, and a characteristic audio signal of a person. Optionally, a trained model is used to detect audio signals. Optionally, the audio signals are used to alter events and / or quality characteristics.

[0116] Optionally, the system comprises a spatial tracker, wherein the processor is configured to track the spatial position of a boat and / or a user of the fishing rod.

[0117] Optionally, the processor is configured to: determine a spatial path from the spatial tracker and associate a spatial path to the one or more events. Optionally, the spatial path comprises a movement of the boat (drift of trolling movement) and / or the user of the fishing rod. Optionally, the processor is configured to determine, based on the spatial path of a plurality of events, an event map. Optionally, the event map comprises an overlay of the spatial path of each of the plurality of events.

[0118] Optionally, the system is configured to output a display comprising: the vehicle movement; and events occurring along the vehicle movement. Optionally, each of one or more events has an event identifier, optionally an icon. Optionally, interaction with the event and / or event identifier is configured to access event data associated with the event.Optionally, one or more events are shown by intensity maps. Optionally intensity maps comprise one or more of: dots, vectors, heatmap, engagement lines, area grids, and linked graphs. Optionally, the vehicle movement is neutrally coloured.

[0119] Optionally, the system comprises the one or more sensors. Optionally, the system comprises one or more cameras as sensors.

[0120] A further disclosure comprises a computer implemented method for monitoring a fishing rod, the method comprising the steps of: receiving an output from one or more sensors; determining one or more regions of at least one fishing rod from the output; and determining one or more events based on the output in the one or more regions of the at least one fishing rod.

[0121] Any one or more of the described options of the system may be applied to the computer implemented method.

[0122] A further disclosure comprises a monitoring system for an item of sports equipment, the monitoring system comprising: a processor configured to: receive an output from one or more sensors; determine one or more regions of at least one item of sports equipment from the output; determine one or more events based on the output in the one or more regions of the at least one item of sports equipment. Optionally, the item of sports equipment comprises one or more fishing rods and / or one or more fishing lines.

[0123] Optionally, the output is obtained from a storage medium. Optionally a hard disk or server. Optionally, the disclosure comprises the method of the system. Optionally any one or more of the described options of the systems herein and above may be applied to this system.

[0124] A further disclosure comprises a system for producing a video summary of an activity, the system comprising: a processor configured to receive video from one or more sensors, determine an event notification based on the received video and save at least a portion of the received video. Optionally the activity is fishing. Optionally the activity is a hunting or sporting activity.

[0125] A further disclosure comprises a system for producing a video summary of an activity, the system comprising: a video storage buffer configured to temporarily store video from one or more cameras; and a processor configured to receive an event notification and save at least a portion of the temporarily stored video. Optionally the activity is fishing. Optionally the activity is a hunting or sporting activity.

[0126] A further disclosure comprises a system for producing a video summary of a fishing activity, the system comprising: a video storage buffer configured to temporarily store videofrom one or more cameras; and a processor configured to receive an event notification and save at least a portion of the temporarily stored video.

[0127] Optionally, the processor is configured to receive event notifications from an event sequence. Optionally, the processor is configured to determine if a received event notification initialises or fits into an event sequence. Optionally, on receiving a first event in the event sequence the processor is configured to save a portion of the temporarily stored video and the processor is configured to continue storing video until a final event is notified.

[0128] Optionally, the processor is configured to save the stored video after a final event in the event sequency is received. Optionally, the processor is configured to continue to store video for a time period after the final event is notified.

[0129] Optionally, event sequences comprise a series of events from an initial event to a final event. Optionally, the processor is configured to determine event sequences algorithmically based on the detected events. Optionally, the processor determines if determined events are initial, intermediate, or final events.

[0130] Optionally, an event sequence is determined by the model. For example, the processor may be trained by labelled training data of event sequences to determine initial and final events and / or event sequences. Optionally, the processor is configured to initialise an event sequence when an initial event is determined. Optionally, the processor is configured to add to the event sequences as intermediate events are detected. Optionally, the processor is configured to end an event sequence when a final event occurs. Optionally, one or more images and / or videos are associated with each event.

[0131] Optionally, the processor is configured to initialise a plurality of separate event sequences concurrently. Optionally, the processor is configured to determine if a determined event is a continuation of a current event sequence. Optionally, the processor is configured to associate the determined event to the current event sequences. Optionally, the determination is based event characteristics. Optionally, event characteristics comprise temporal and / or spatial proximity. Optionally, the processor is configured to determine if the determined event is a final event.

[0132] The system may apply any of the event sequence processes disclosed for this or other systems and methods.

[0133] Optionally, the processor is configured to receive a quality characteristic associated with the event.The system may apply any of the quality characteristic processes disclosed for this or other systems and methods.

[0134] Optionally, the processor is configured to adjust a characteristic of the portion of the stored video based on a quality characteristic. Optionally, the characteristic is one or more of a length, saved time period, quality, and / or type of stored video.

[0135] Optionally, the processor is configured to save, during an event sequence, ongoing video. Optionally, the processor is configured to save the ongoing video at least until receipt of a future event notification (event end notification). Optionally, the processor is configured to delete ongoing video based on the quality characteristic.

[0136] Optionally, the video storage buffer is configured to store less than 120 seconds of video, less than 60 seconds of video, or less than 30 seconds of video. Optionally, the video storage buffer comprises a rolling storage buffer.

[0137] Optionally, the processor is configured to determine a portion of the temporarily stored video to save based on the event notification and / or based on determining one or more events occurring in the temporarily stored video.

[0138] Optionally, the processor is configured to combine a plurality of saved videos. The saved videos may be from events in an event sequence.

[0139] Optionally, the processor is configured to receive event notifications and determine a status category of the event notification. Optionally, the status categories comprise: set-up, fight, and resolution status. Optionally, the processor is configured to save videos comprising videos from two or more status categories, or each status category.

[0140] Optionally, the processor is configured to produce a summary video. Optionally, a deterministic algorithm is configured to select a set of events in the event sequence.

[0141] Optionally, events are selected based on quality characteristic. Optionally, the processor is configured to order the events in time. Optionally, the processor is configured to combine video associated with the events into a single video (highlight video). Optionally, the video comprises narrative sequence.

[0142] Optionally, the processor is configured to add narrative or context video or images to the highlight video. Optionally, the context video or images are received from an additional sensor. Optionally, the context video or images are provided by user input.

[0143] Optionally, the processor is configured to receive video and event notifications from a plurality of sensors. Optionally, a primary sensor provides the received video and at least one secondary sensor provides additional video and / or images.Optionally, the video storage buffer discards temporarily stored video if no event is detected.

[0144] A further disclosure comprises a system for producing a video summary of a fishing activity comprising a processor configured to:

[0145] receive an event notification and identifying an associated video;

[0146] determine if the event notification is an initial, intermediate or final event in an event sequence and:

[0147] if the event is an initial event initialise a video summary for the event sequence with the video;

[0148] if the event is an intermediate event, add the video to the video summary of the event sequence; and

[0149] if the event is a final event, add the video to the video summary of the event sequence and ending the event sequence and / or video.

[0150] A further disclosure comprises a computer implemented method for producing a video summary of a fishing activity, the method comprising:

[0151] receiving an event notification and identifying an associated video;

[0152] determining if the event notification is an initial, intermediate or final event in an event sequence and:

[0153] if the event is an initial event initialising a video summary for the event sequence with the video;

[0154] if the event is an intermediate event adding the video to the video summary of the event sequence; and

[0155] if the event is a final event, adding the video to the video summary of the event sequence and ending the event sequence and / or video.

[0156] The steps of the above systems and methods may be optionally applied to this system and method.

[0157] Optionally, the system or method are configured to receive event notifications and identify associated media, such as images and / or audio. Optionally the system or method are configured to determine if the event is an initial, intermediate of final event using a model.

[0158] A further disclosure comprises a computer implemented method for producing a video summary of a fishing activity, the method comprising:

[0159] storing received video from one or more cameras in a video storage buffer; receiving an event notification; andsave at least a portion of the temporarily stored video.

[0160] Optional features of the systems disclosed herein may be applied to the method. As used herein the term “and / or” means “and” or “or”, or both. As used herein “(s)” following a noun means the plural and / or singular forms of the noun. For the purpose of this specification, where method steps are described in sequence, the sequence does not necessarily mean that the steps are to be chronologically ordered in that sequence, unless there is no other logical manner of interpreting the sequence. The term “comprising” as used in the specification and claims means “consisting at least in part of.” When interpreting each statement in this specification that includes the term “comprising,” features other than that or those prefaced by the term may also be present. Related terms “comprise” and “comprises” are to be interpreted in the same manner.

[0161] This invention may also be said broadly to consist in the parts, elements and features referred to or indicated in the specification of the application, individually or collectively, and any or all combinations of any two or more said parts, elements or features, and where specific integers are mentioned herein which have known equivalents in the art to which this invention relates, such known equivalents are deemed to be incorporated herein as if individually set forth.

[0162] BRIEF DESCRIPTION OF THE DRAWINGS

[0163] One or more embodiments of the invention will be described below by way of example only, and without intending to be limiting, with reference to the following drawings, in which:

[0164] is a schematic of the coverage to be provided by a fishing system.

[0165] is a schematic of the value offered by a fishing system.

[0166] is a block diagram showing the connectivity between components and data sources according to one embodiment of a fishing system.

[0167] is a block diagram showing the connectivity between components and data sources according to another embodiment of a fishing system.

[0168] is a block diagram showing the connectivity between components and data sources according to another embodiment of a fishing system.

[0169]

[0170] 3D is a block diagram showing the dual-pathway architecture for asynchronous, post-trip processing of historical or third-party media according to another embodiment of a fishing system.

[0171] shows a sensor module and its components for a fishing system.shows an exemplary set up of a fishing system on a boat.

[0172] shows an exemplary set up and use of a fishing system on a boat. shows an exemplary set up of sensor for a fishing system on a boat. shows an exemplary set up of sensor for a fishing system a boat. shows an exemplary set up of sensor for a fishing system on a boat. shows an exemplary set up of sensor for a fishing system with indications of the fishing rod regions.

[0173] shows a flow diagram for event detection using the fishing system. shows a flow diagram of video processing based on event detection and / or event sequence detection.

[0174] shows an example architecture of the multi-path data processing architecture and event-driven rolling video buffer logic.

[0175] shows a flow diagram for training data in a two-model system.

[0176] shows a flow diagram for training data in a one-model system.

[0177] shows an example spatial map generated by the fishing system.

[0178] DETAILED DESCRIPTION OF EMBODIMENTS

[0179] Throughout the Figures, like references refer to like components.

[0180] Current systems are not suitable for recreational anglers and do not offer real-time, automated analytics of, for example fishing performance.

[0181] Optionally, activities such as cycling, running, motorsport, golf, and skiing have systems to provide real-time tracking and performance monitoring to enhance user experience and performance. For example, in cycling, GPS-enabled cycling computers and mobile apps like Strava™ and Garmin™ allow riders to automatically log data such as speed, distance, and elevation during the on-bike experience that provide instant insights into their metrics. These may enable subsequent off the bike experience analysis for a wholistic complete cycling experience. Similarly, in running, devices like watches and applications such as the ASICS Runkeeper™ app offer real-time tracking of pace, distance, and heart rate, helping runners improve their technique and performance. Similarly, motorsports use telemetry system to collect data from vehicles during races to track engine performance, tyre pressure and driver performance in real-time. However, these systems simply provide performance tracking of measurable performance. Moreover, they are unsuitable to provide monitoring for complex activities. Complex events may include, for example, where usersmay interact with an object, an elongate object, a flexible object and / or multiple objects and / or where a range of events must be monitored to determined performance.

[0182] Fishing provides a particularly complex environment because a user may interact with a range of different fishing rods and equipment (such as lines and lures). Fishing rod movements are complex. Useful performance metrics cannot simply monitor the use of the fishing rod, but must determine the actions of the fishing rod, and separate other movements on the boat. There is not a single performance metric to monitor (such as distance or time). Instead fishing performance is determined by a series of events and may require determining how the events are linked. Fishing systems do not operate in a stable environment and are not human centric. Sensors based on human activity omit important fishing events. Using a passive sensor (such as an image, video, range sensor) targeting the fishing gear (such as the fishing rod) ensures fishing events are captured.

[0183] It is possible to provide automated tracking systems for fishing by applying sensors to fishing rods. However, this adds weight to the fishing rod altering the balance and performance of the fishing rod and doesn’t consider the use case of users who utilise multiple rods. Manual input of information may be required to record the catch details such as the type and size of a caught fish.

[0184] There are currently no existing fishing systems that usefully track on the water fishing performance. It is advantageous to provide a passive system to track casts, catches, and fight performance without, or with reduced, user intervention.

[0185] Users may want to display their fishing performance. This may require images, or desirably video. In fishing competitions, it may be advantageous to have a video record of fishing performance. While manual preparation of fishing highlight videos and / or fishing competitions is possible these systems require substantial time and / or people on the boats. It is advantageous for a fishing monitoring system to provide an output suitable for video preparation and / or for a video processing system to be configured for preparing a video summary of fishing events and / or fishing performance.

[0186] Particular advantages may be found in Bass Fishing. Bass fishing is typically performed on small boats with tens or hundreds of casts per hour. The system provides a method to capture events for each of the casts from the Bass boat. Optionally, the system also allows those events to be used to generate a record or highlight video of the fishing. The highlight video may concentrate on key events, reducing the length of the video substantially. This may be applied in fishing contests, where the record video shows fishes caught and / orfishing metrics, without requiring review of the entire video and / or capture of fish. In some cases, the video and / or highlight video may be provided to a cloud and / or server for streaming. In some cases, the upload and / or streaming is automated. In some cases, a quality characteristic may be used to determine if streaming occurs. In some cases, an event notification may trigger a livestream.

[0187] In one example, the present disclosure shows a fishing system. The fishing system may be referred to as a monitoring system for at least one fishing rod. The monitoring system comprises processor configured to: receive an output from the one or more sensors; determine one or more regions of at least one fishing rod from the output; and determine one or more events based on the output in the one or more regions of the at least one fishing rod. The one or more sensors configured to monitor an area including at least one fishing rod may form part of the system.

[0188] Advantageously the identification of one or more regions of the fishing rod, allows fishing specific events to be determined from a sensor monitoring the general area of the fishing rod. This can remove the requirement to have sensors on the fishing rod, optionally reducing the number of sensors and the need to move sensors between equipment. In some examples, the present disclosure describes the analysis of a fishing rod in a plurality of regions of the fishing rod. By identifying specific regions of the fishing rod, the determination of the events can be more accurate than attempting to determine events from the entire fishing rod. This is because the overall fishing rod can look similar when different events occur, but the high frequency and / or detailed movements in regions of the fishing rod can distinguish between these events.

[0189] In a fishing system the regions may comprise one or more of a rod tip zone; a midshaft zone; a butt zone; a line zone, and a reel zone. The selection of these zones provides regions of the fishing rod which act differently, or which may allow specific events to be more easily detected. For example, changes in reel handle rotation and spool motion can be analysed alongside the rod and line zones when determining events such as winding, fighting a fish, or retrieving slack. The system may analyse the overall shape or motion of the entire fishing rod as a single object. However, defining separate regions is advantageous because different fishing events produce distinct patterns in each region. This allows a more accurate classification than treating the whole rod as one undifferentiated region and may also reduce the processing power required. One or more regions may be defined for additional fishinggear, such as gear interacting with the fishing rod or fishing line, outriggers, downriggers, catch storage and rod storage.

[0190] The analysis of a fishing rod may comprise determining a position and / or movement of the fishing rod and / or the fishing line. In some cases, the determination of an event comprises determining a position and / or movement of the fishing rod in the one or more regions. However, in some cases, the determination of the position and / or movement of the fishing rod may be implicit. In such cases the output of the system may be a fishing event without detail of the position and / or movement of the fishing rod provided as a defined output. Advantageously the use of regions focusses the model on relevant areas of the fishing rod. In some cases, the fishing system may determine events based on the fishing line and regions of the fishing line. This may be without use of the fishing rod.

[0191] The system may use any known or discovered positions and / or movements. Examples include: tip twitch, shaft deflection, and / or tension variations in the line (slack - taut). In some cases, a filter is applied to focus on high-frequency motion such as line jumps, twitches or lateral movement. The filter may be a low pass filter. The selection of a suitable high-frequency threshold may depend on the dominant motion of the vehicle. This may be determined for each session or predetermined. In some cases, a threshold frequency is chosen to separate rod-tip vibrations from boat movement. In some cases, the dominant motion of the hull in that session is determined by a frequency analysis of a session and filtering below a dominant low frequency. In some cases, position and / or movement comprise: rotational (e.g., the motion of reel components such as the handle / spool), rates of change of curvature (of the fishing rod), persistence of bend patterns; curvature changes along the rod shaft and / or changes in the tension of the fishing line.

[0192] Determining an event may be based on the movement and / or position of the fishing rod s over a period of time. Advantageously using a period of time reduces the reliance on a single image. The processor can use changes in the images to determine movement, to determine tension or changes in tension, and / or to link movement in different regions. This additional information improves event detection and extends the types of events which can be detected. In some cases, tracking regions of the fishing rod over time allows the system to detect specific visual signatures - such as progressive shaft deflection, sudden transitions between a slack and taut line, or lateral movement of the line before tension reaches the rod.

[0193] In some cases, the system is configured to detect events directly from the behaviour of the fishing rod (including the line), and / or the fishing line). This allows the system tooperate even when the angler is partially or wholly outside the frame, when rods are unattended in holders or bank spikes, and when partial views of the fishing rod are available. The system may use the time-varying physical deformation of the rod and line as the primary input signal to a processor for event detection (and optionally automated video buffer control). Any observable human body motion may be used as only an optional secondary cue.

[0194] The system may use a range of sensors. Example sensors include image sensors, video sensors (cameras) and range sensors (radar, Lidar, mmWave, etc). The sensors are arranged with their field of view covering an area where fishing is expected to occur. For example, the area may be for deck of a boat, or the bank of a body of water. The field of view may cover substantially all the area where fishing is expected to occur, or a portion thereof. In some cases, multiple sensors may be used and / or areas monitored.

[0195] In some cases, one sensor location is raised (for example on a pole near the console or stern of a boat). The sensor may be positioned to look at the fishing rod from behind. Most of the bow or deck of the boat may be in view. In some cases, the sensor may have a 360-degree view. In some cases, the sensor may be above the fishing rod, or at least the reel of the fishing rod.

[0196] In some cases, the field of view of the (first) sensor is divided into one or mode layout regions. In some cases, the one or more layout regions each including at least one fishing rod position. In some cases, the one or more layout regions each include the path of the fishing road position to where a line of the fishing rod in the fishing rod position would meet the water. In some cases, the system is configured to identify individual fishing rods in the field of view of the sensor.

[0197] In some cases, a sensor location is on the bow of a boat. The sensor location may be at or near the level of the deck of the boat. The sensor may be positioned to view the fishing rod from the front. The sensor may be positioned below the expected position of the fishing rod. The sensor may identify one or more layout regions, each including at least one fishing rod position. In some cases, the layout regions each include one or more regions of the fishing rod and one or more regions including at least a portion of the line path from the rod tip down towards the water in front of the boat. In some cases, the layout regions each include a region about the angler’s upper body.

[0198] In some cases, a sensor location is on the holder of the fishing rod and / or on the fishing rod. In some cases, the sensor location is a point-of-view sensor, a chest mounted and / or a hat mounted sensor. The sensor may identify one or more layout regions, eachincluding at least one fishing rod position. In some cases, the sensor determines a single fishing rod position. In some cases, the single fishing rod position is determined as the fishing rod associated with a user of the sensor. In some cases, the layout regions each include one or more regions of the fishing rod. In some cases, the layout regions include the rod tip and at least a portion of the line between the rod tip and the water. In some cases, the sensor is used in combination with the other sensors. In some cases, the sensor provides outputs associated with fine detail on hook sets and fights.

[0199] In some cases, training data is provided from multiple sensor positions. In some cases, sensor positions comprise one of more of the stern of the boat; the bow of the boat; body worn; mounted to the fishing rod; and external to the boat. The sensors field-of-view will depend on their placement on, for example, a boat. The sensor may have a layout region. The layout region refers to a repeatable area of the sensor view. A sensor may have multiple layout regions. For example, the sensor may have separate layout regions for each fishing rod on a boat, or each active fishing rod on a boat. Using layout regions may simplify the processing of sensor output because multiple rods do not need disambiguation in the sensor output. The system may predefine layout regions and / or the system may determine layout regions, for example by computer-vision models to identify fishing rods.

[0200] In some cases, the processor receives outputs from one or more data sources. The data sources may comprise additional sensors, prior data and / or user inputs. For example, the processor may receive data on the size or weight of a fish, or an image of the fish after a catch event and associate this with the event. The processor may receive audio information associated with the event and / or a note from a user about the event. The data sources may also be used to determine events. For example, audio input (such as speech) may be used to determine events. For example, a picture taken on a camera (such as a camera for fish measurement) may mean the processor determines a catch event. In some cases, this may be added to detected events and / or as new detected events if the event was obscured from the primary sensor. In some cases, the processing depends on the current event sequence.

[0201] In some cases, the processor is configured to determine vessel(vehicle) movement based on the output of a vessel position sensor and stabilise and / or modify the output of the sensor. This may address complexities of operating on a boat. A boat, is typically a six-axis moving platform (pitch, roll, yaw, heave, surge, sway) in front of a constantly changing background of waves, glare, clouds, and shifting light. Coupling video analysis to inertial measurement unit (IMU) data so that vessel motion is estimated in real time and subtractedfrom the video stream, isolating higher-frequency geometric deflection of fishing gear such as a rod tip or moving line against the moving water.

[0202] In some cases, the processor is configured to determine fishing gear associated with the fishing rod. Associated with the fishing rod may include attached or connected or touching the fishing rod and / or the fishing line. Examples, include outriggers, downriggers and rod supports (such as rocket launchers). Outriggers and / or downriggers and other devices which have a level of flexibility may include one or more regions. In some cases, the fishing line region may be divided based on the fishing gear. For example, a first fishing line region may span between the rod and the fishing gear (e.g., an outrigger) and between the fishing gear and the water surface. In some cases, an adjusted or different model is used if fishing gear, such as an outrigger or downrigger is used. The adjusted model may add additional regions for the fishing gear and / or have additional training data.

[0203] In some cases, the processor is configured to estimate the gross motion of the vessel (for example including pitch, roll and yaw components) from the IMU data over time. The processor may compensate for that motion when analysing the fishing rod regions.

[0204] Advantageously this ego-motion compensation may improve discrimination between apparent motion caused by waves or vessel movement and true geometric deformation of the rods and lines, thereby reducing false positives and allowing the system to resolve small deflections, such as a subtle bite or tip twitch, on a moving platform.

[0205] In one example, compensation is achieved by applying a kinematic coordinate transformation that subtracts the calculated 6-axis vessel motion vector from the raw optical tracking data in real-time. The compensation may be achieved by applying a kinematic coordinate transformation that subtracts the calculated 6-axis vessel motion vector from the raw optical tracking data in real-time.

[0206] The present disclosure describes the use of deterministic models (such as neural networks, such as convolutional neural networks (CNN)) and / or machine learning models to determine the one or more events. These may be referred to as models and / or Al. Models may be machine learning models and / or artificial intelligence models trained on suitable training data. Example machine learning models are Large Language Models (LLMs), VLMs (Vision-Language models) and computer vision models. Models may be multi-modal. Other frameworks (including generative frameworks) capable of processing video or audio frames may be used. The output of the models may comprise event determination, or semantic understanding for editing and analytics. The disclosure may be implemented with anysuitable model and provided with the training data. In some cases, analytical methods are used. Analytic methods may be based, for example on kinematics in the one or more regions. In some cases, model development continues in use. Data collected during use may be used as additional training data to update the model. User input may be used to adjust and / or correct models.

[0207] Although trained models may be used the system could operate with a computervision system or pipeline.

[0208] In some cases, the system uses deterministic computer-vision pipelines. In one example one or more vision models analyse the sensor outputs (images and / or video frames) to detect fishing rods and track the movement and / or position of the rods and / or regions of the rods. Additional tracking may include anglers, fish, and the surrounding environment if available. The models may determine features such as rod curvature, rates of change in curvature, persistence of bend patterns, line slack and / or tension states. These features may then be used to determine event.. Example events include no-load, transient impulse, sustained load, and landed or lost fish. Events may be determined based on the features by, for example, a classifier or explicit rule-based state machine. In alternative cases, the events are produced by end-to-end trained models, such as learned or multimodal Al models that operate directly on the sensor input(s).

[0209] In some cases, the processor is configured to apply a first model configured to determine the one or more regions of the at least one fishing rod and a second model configured to determine the one or more events based on the determined the one or more regions of the first model. In some cases, the first model comprises a determinative model. The first model may be a computer vision pipeline or convolutional neural network. In some cases, the second model comprises a machine learning model. The first model may be considered a preprocessing stage.

[0210] It may be advantageous to have a first model for identifying the one or more regions because this may be a relatively straightforward and can be separated from the more complex problem of event identification. The use of the first model may reduce the size and / or complexity of the overall process. In some cases, this allows the system to be implemented on a smaller processor, such as an edge device. The use of two models may also simplify the training of the models because training images can be specific to regions of the fishing rod. The separation of regions before identifying movements also helps target the model on key regions of the fishing rod, reducing errors based on, for example, extraneous movements.T1

[0211] In some cases, the first model is trained to determine a first of the one or more regions and determine at least a second of the one or more regions based on the location of the first region.

[0212] In some cases, a model is trained to determine the one or more events from the received outputs. The model may not provide external access to the one or more regions determined by the model. In some cases, the model is trained on training data comprising a plurality of images labelled with the one or more regions and the one or more events. In some cases, the model is trained on training data comprising a plurality of images labelled with the one or more events, only the one or more events, or without the one or more regions of the fishing rod.

[0213] The models are trained by training data. The training data may comprise one or more pluralities of images (datasets). Training data may comprise one or more datasets in which images have been labelled (semantically tagged). At least one dataset may include images of fishing rods where at least one, or all, of the one or more regions have been labelled. In some cases, the regions are labelled as an area of the image. In some cases, the images are labelled with a plurality of different regions.

[0214] In some cases, the images may include a plurality of fishing rods. The images may be divided to form a plurality of images, each showing one fishing rod. The images may be labelled to identify a plurality of the regions on each of the plurality of fishing rods. The type of model used may determine the most effective processing of multiple fishing rods.

[0215] In some cases, the training data comprises a plurality, preferably at least thousands, of images. In some cases, the training dataset comprises time series of images, such as videos, photo-bursts, and / or series of images. In some cases, the images comprise frames from videos. In some cases, the images, or at least a portion of the images, are time-stamped. In some cases, the images are labelled with one, or both of, at least one region and at least one event. In some cases, the training data comprises a plurality of videos. In some cases, the training data are from one or more sensors positions, or each known sensor position. This can improve model training by avoiding confusion between rod movement in different three-dimensional positions. In some cases, the sensor positions are correlated with, or the same as, the sensor positions used in the system.

[0216] In some cases, the processor is configured to determine active fishing rods. In some cases, only fishing rods determined as active are processed by the processor. Determining whether a fishing rod is active or inactive (i.e. a stored rod for example) may be based onrecent motion, orientation, and position in known storage locations such as fishing rod frames (for example rocket launchers) or cabin racks). In some cases, determination of an active fishing rod will determine a layout region. In some cases, the processor is configured to determine active fishing rods in each layout region.

[0217] Use of the systems and / or models may introduce additional training data. Images captured during use may be used to train the models and / or system. User input may be provided and / or requested, for example if a user notices an error or the system prompts for clarification. In some cases, the additional training data is labelled automatically based on region and / or event detection from associated images.

[0218] In some cases, the fishing rod may, at least in part, be occluded from the field of view of the sensor. For example, a user may move between the fishing rod and the sensor. The system may be trained to, where information is not available on a region of interest, to use other regions. In some cases, the training data comprises a plurality of images with occluded regions of the fishing rods. In some cases, the training data comprises one or more series of images in which one or more of the regions are occluded for only a portion of the series. Advantageously, this allows the model(s) to train to infer an event even when regions are occluded. In an alternative the processor is configured to determine if one or more of the regions is occluded and determine the event without use of the occluded region.

[0219] In some cases, the outputs from the sensor comprise a time characteristic. The time characteristic may allow a time-based comparison between outputs and / or regions.

[0220] In some cases, the one or more events comprise a time characteristic. The time characteristic may allow additional sensor data to be associated with the one or more events. A time characteristic may refer to, for example, a time stamp, a start time, and end time and / or a time period associated with the output and / or event. In some cases, the event time characteristic is determined based on one or more time characteristics of the outputs.

[0221] In some cases, a processor is configured to determine event sequences. Event sequences comprise a series of events from an initial event to a final event. In fishing systems, the event sequences typically reflect a fishing action (such as from cast to release). The models may be trained to determine event sequences. In some cases, event sequences are determined algorithmically based on the detected events, for example by considering if determined events are initial, intermediate, or final events. Advantageously event sequences provide more information to the system. This may allow improved decision making on howto react to an event, such as how much video footage to save. Event sequences also improve the analysis and metrics as they link the outcomes of events to the initialisation of the events.

[0222] The system may track a quality characteristic of an event or event sequence. The quality characteristic may, for example consider an excitement of the event or event sequence and / or an importance of the event or event sequence. For example, if greater excitement is determined the event may be more important, or if a larger fish is captured the event may be more important. The processor may process the event based, in part, on the quality characteristic. In some cases, the processor may determine an amount of images and / or video to save based on the quality characteristic. In one example a numerical value is updated during an event sequence, with the output and any additional sensor output used to adjust the numerical value based on determined quality.

[0223] Sensors may be mounted to a person, a vehicle (such as a boat), fishing equipment and / or the ground. The sensors may be moveable or fixed (such as in a fixed holder, bank spike, rocket launcher). In some cases, the system operates on a personal electronic device connected to the sensor. This allows use in a variety of situations. Advantageously, mounting the sensor to a fishing boat reduces the set-up required for the system because it does not have to be adjusted for particular users and equipment.

[0224] Boats will move due to water movement. Advantageously the system accounts for this motion when determining events. Stabilising the output of the sensor for vehicle movement allows the line deformation to be measured consistently in each of the one or more regions. The boat movement output can be used to adjust or qualify movements detected in each of the regions. For example, certain drift, heave or surge patterns caused by vehicle movement may assist the movement of fishing rods (particularly when in holders) in holders to work a lure. In some cases, the low frequency movement may be adjusted for, as this is likely movement of the boat rather than movement of the fishing line.

[0225] In one example, the present disclosure shows a fishing system output. The output may be displayed on a screen. The output is a spatial path of one or more fishing events. The output may be created by a processor: determining a spatial path from the spatial tracker and associating a spatial path to the one or more events. Advantageous associating a spatial path provides a history of the movement leading up to the event. This is useful in fishing because the movement before the catching of a fish is often more relevant than where the fish was finally caught. The output may comprise a path of the boat (or fishing rod) with events and / or event sequences marked on the path. The events and / or event sequences may be selectable toprovide further details. Advantageously this allows a user to clearly see the events and outcomes displayed relative to the spatial path.

[0226] In one example, the present disclosure shows a system for producing a video summary of a fishing activity. The system is configured to temporarily store video from one or more cameras in a video storage buffer. When an event notification is received a processor saves at least a portion of the temporarily stored video. Advantageously this reduces the amount of video storage required because video is only stored based on event notifications. The event notifications may come from the fishing system.

[0227] In some cases, the processor is configured to receive notifications from events in an event sequence. The processor may modify the storage of video based on each of the event notifications. For example, the first event may initial video storage, including from the video buffer, while intermediate events (or the timing of intermediate events) may increase or decrease the amount of video stored (for example if there is a long time between cast and bite the video may be cut).

[0228] In some cases, the system for producing a video summary is configured to perform the event sequence and / or quality characteristic steps described herein. In some cases, the fishing system and video summary system are a single system, or linked systems. In some cases any one or more steps is performed by either system. Advantageously having the video summary system configured to work on event sequences allows automated highlight reels to be prepared which show complete event sequences and may also clip or modify video to focus on key highlights. For example, an event sequence may be cast, bite, catch. Each event may be passed to, or determined by, the system. The system may initialise an event sequence with the cast (an initial event). A video of the event may be saved. A buffer may be used to ensure full coverage pre and post event in the video. The event sequence may remain open awaiting a next event (in a predefined series, or as determined by the system) The bite (an intermediate event) may then be added to the initial video once detected. The catch (a final event) may trigger the end of the event sequence, with a final video added to the video sequence. In some cases, once an initial event occurs, video may be recorded until the final event. In some cases, based on event characteristics (such as timing, quality characteristics, additional sensors), the amount of video recorded over the event sequence (or quality of the video) may be modified and / or portions (for example of low activity) may be deleted.

[0229] In some cases, the processor is configured to add narrative or context video or images to the highlight video. Shots such as the angler talking to camera, scene setting footage, ortravel between spots and alternative camera views are also tagged by time and can be interleaved wherever they overlap with, or logically sit between, those event chains, similar to how sports highlights cut between angles and sideline commentary.

[0230] A fishing system 100 of the current invention is described herein with reference to Figures 1 to 15.

[0231] In some cases, the fishing system 100 is a non-intrusive, vision-first artificial intelligence (Al) system that functions primarily as an event-driven, automated media production and / or trip-analytics engine. While the system 100 may automatically track fishing performance and boating analytics, one feature is to utilize computer vision and sensor fusion to detect fishing-related events, such as casts, bites, strikes, missed strikes, fights and landed fish. In some cases, these detected events act as system 100 triggers (also referred to as event notifications) used to manage video production. The video production may include a continuous rolling video buffer. The video production may automatically extract and semantically edit video highlight clips and / or high-resolution still images. In one example the video production assembles a comprehensive 'Trip Report.' The Trip Report is a unified composite artifact combining curated media with aggregated performance metrics and environmental summaries. Accordingly, per-event logs and statistics generated by the system may serve primarily as supporting data structures to index, annotate, and summarize media within the broader narrative of the Trip Report, rather than functioning merely as standalone event logs. Video highlights or a video summary may form part of the trip report.

[0232] The fishing system 100 may use the continuous rolling video buffer (for example, 30 to 60 seconds in length) across one or more cameras capturing the environment. When a fishing event indication is received (such as a strike, cast, or landed fish) the system 100 may retroactively save an event segment from this buffer. The event segment may by a configurable pre- and post- amount or may be dependent based on the event, an external sensor and / or a quality characteristic. This pre-emptive capture ensures that the critical setup and moment of action are preserved without requiring always-on recording. The system 100 may dynamically apply different retention policies based on the event type (e.g., saving a longer segment for a landed fish versus a shorter segment for a missed strike).

[0233] In some cases, the system further aggregates the event locations into spatial cast-density and effort maps. For each detected cast, strike, fight, landed fish, or missed strike, the system records the corresponding vessel position and, where available, chart or bathymetric data describing nearby geographic or underwater features such as shorelines,weed lines, drop-offs, channels, and submerged structure. In some cases, the system continuously records a vessel heading or orientation signal, such as compass heading or course over ground, and associates this orientation with each detected fishing event so that cast directions, rod positions, and lure or bait deployment lanes are interpreted relative to both the boat and the surrounding geography. Over the course of a trip or across multiple trips, the processor is configured to generate heatmaps and other spatial visualisations that represent the concentration and distribution of lure or bait deployments relative to the vessel track, vessel orientation,

[0234] In some cases, the system further models rod spreads and per-rod performance within the same spatial framework. For each detected fishing event, the processor associates the event not only with a geographic location and vessel orientation but also with a specific rod position within a predefined rod layout relative to the vessel, such as port and starboard gunwale positions, transom rod holders, outriggers, or bank spikes. Over time, the system accumulates statistics per rod position, including rates of strikes, hookups, landed fish, and missed strikes, and correlates these with lure or bait types, depths, and casting or trolling angles. The resulting cast-density and rod-spread analytics allow users to visualise which water lanes and rod positions around the vessel are most productive under given conditions, to identify under-fished sectors, and to adjust their spread and casting patterns accordingly.

[0235] In a first example the fishing system 100 may comprise at least one non-contact sensor (such as a range or an image visual sensor) and a processor (such as an Al-driven analytics processor) to provide a data capture experience. In one example the non-contact sensor is a millimetre-wave (mmWave) radar configured to detect an object’s position, distance, and / or velocity. The object may be a fishing lure, fishing rod, a fish, a user or other similar object. The fishing system 100 may track casting motions, rod activity, and fish fighting dynamics.

[0236] Herein a sensor may comprise a visual sensor. A visual sensor may comprise one or more image sensors configured for multi-spectrum optical capture. As used herein, the terms 'camera', 'video', and 'visual sensor' are not limited to standard continuous frame-rate video or the visible light spectrum. These terms explicitly encompass data generated by any image sensor capturing sequential still images, low-framerate optical bursts, or continuous video streams across multiple wavelengths, including the visible spectrum, thermal imaging, infrared (IR), and specialized low-light or night-vision spectrums. The definition of 'vision', ‘video’, ‘visual sensors’ and 'camera' explicitly encompasses the visible light spectrum,thermal imaging, infrared (IR), and low-light night-vision spectrums, ensuring continuous event detection across these zones regardless of optical hardware or lighting conditions. In various embodiments, the visual data (images) may be provided by any image sensor configured to capture still images, sequential still images, low-framerate optical bursts, or continuous streams across one or more wavelengths, including visible, thermal, infrared, or low-light and night-vision bands.

[0237] Furthermore, a processor (which may be configured with Al-driven analytics) is configured to ingest and analyse outputs from the visual sensor to provide continuous event detection regardless of lighting conditions or hardware format.

[0238] The term 'Al' as used herein encompasses traditional deterministic machine learning models (e.g., Convolutional Neural Networks) as well as multi-modal Large Language Models (LLMs), Vision-Language Models (VLMs), and generative frameworks capable of processing video or audio frames to provide semantic understanding for editing and analytics.

[0239] In some examples, further sensors may be used to provide environmental context, such as light sensors, temperature sensors, barometers, inertial measurement units (IMUs), GPS-location sensors, additional visual sensors, audio sensors and acoustic or sonar sensors. User data may be input into the system 100 and / or stored data may be available to the system 100.

[0240] Figure 3 A shows an example of system 100. Conventional systems may solely query location coordinates at the isolated moment a fishing event occurs. In contrast some examples of the system 100 utilize continuous spatial tracking to capture the broader movement (e.g. kinematic context) of the vessel. Location data may be continuously ingested. For example, via an integrated Global Positioning System (GPS) receiver within a sensor module 50, or by wirelessly receiving continuous NMEA 2000 telemetry from the vessel’s marine electronics or a coupled smartphone. This continuous tracking allows the system 100 to analyse the vessel's drift vector, trolling speed, and historical path leading up to a determined event.

[0241] In some cases, when a fishing event (e.g., a strike) is detected, the system 100 does not merely log a static waypoint. In some cases, system 100 correlates the event with the continuous GPS track. This may enable post-trip analytics to reconstruct the exact drifting or trolling technique that resulted in the strike. In some examples, this rich spatial context is subsequently embedded as metadata into the generated trip reports and / or video highlight overlays.As illustrated in figures 3 A to 3D a fishing system 100 according to examples of the invention may comprise a range of components.

[0242] Figure 3 A illustrates an independent or standalone fishing system 100 comprising a power source (e.g. battery) 20, processor 30 (which may be referred to as a data processing unit), a sound module 40; and a sensor module 50. Sensor module 50 may have one or more sensors 500, such as a camera, and one or more additional sensors 501, such as a GPS sensor. The processor 30 may be configured to process data received from the sound module 40 and sensor module 50. While the term processor 30 is used this may be understood to refer to one or more processors with data transfer possible between them. Figure 3 A shows separate video processors 301, core processors, 302 and model training processors 303 but other variations are possible. The fishing system 100 may be configured to couple to a computer 10 and / or a display such as display unit 60. The computer 10 and / or the system 100 may connect to the internet or data in a cloud 80 to enable connection to additional external data inputs 70, such as to weather forecasts, user data, and or water and / or wave forecast. The cloud 80 may also provide connection for sharing data (social media and / or live streaming, and or to providing information and / or fishing performance remotely. The computer 10 and / or the system 100 may be connectable to storage, such as local / home network storage when available. The system 100 may have dedicated storage, at least for operating models and / or storing video. The system 100 may store data in cloud 80. In some cases, the system 100 connects directly to the internet or data in a cloud 80. In some cases, the system 100 is configured to be updatable from cloud 80 through connection to an update server or service 45. The update server 45 may provide updates to one or more models on the system 100 and / or receive updates or additional training data from the system 100.

[0243] Figure 3A shows the system 100 having an internal edge data processing unit 303, a sound module 40, an integrated sensor module 50, and flexible power capabilities (boat power 20 and / or external battery 64). While deployed on a vessel, the system 100 may be connected directly to the vessel's primary power source 20. For off-vessel environments or where boat power is not available (such as kayak fishing, onshore bank fishing), the system is configured to couple to an external power source 64. In some cases, the system comprises an internal power source (e.g., an integrated battery). The internal power source may be configured to facilitate post-trip processing, offline data management, and media offloading.

[0244] In the system of Figure 3 A the internal edge processing unit 303 is configured to process raw visual and sensor data locally to provide low latency event detection withoutrequiring an internet connection. To access cloud infrastructures 80 (for example for pushing automated trip reports or pulling external environmental data), the system 100 may be configured to wirelessly couple to an internet-connected external user device 55 (such as a smartphone utilizing a cellular LTE / SG network) or directly to a local area network (e.g., vessel Wi-Fi or WLAN). In some forms, the system 100 may additionally couple to supplementary external inputs, such as cameras 62, third-party action cameras or wireless scales.

[0245] The cloud 80 connectivity may utilize any known or future wireless communication system, including but not limited to Wi-Fi, satellite, or mobile data networks such as Long Term Evolution (LTE), 4G, 5G, or equivalents. The cloud 80 connectivity may facilitate notifications, live streaming of data, social media uploads and other information delivery or receiving activities.

[0246] In some examples, the fishing system 100 processes all sensor data locally. In some cases, the system 100 uses edge-based computing. Edge computing 303 may provide instant analytics, low latency, and offline capability.

[0247] In some examples, events such as rod casting, fish bites, fights (e.g. reeling in a fish), landing a fish and release or harvesting the fish are automatically detected. The events may be logged. In some cases, statistics are associated with the events. The statistics may be obtained from system 100 or from additional sensors (e.g. camera 62 or other processors).

[0248] The fishing system 100 may be configured to generate a trip record. The tip record may, for example, aggregate environmental conditions, fishing performance metrics, and automatically edited video highlights. The trip record, or portions thereof, may be stored (e.g., in historical records) to use in later analysis and trip planning. In some forms the fishing system may synchronise seamlessly with smartphones, marine electronics, and cloud-based analytics, allowing users to control, review, compare, and share performance insights across devices.

[0249] The system 100 may be configured to couple to additional modules or devices such as a camera 62 and / or an external battery module 64.

[0250] Figure 3B shows a similar fishing system 100 as an exemplary fully integrated system. The system 100 comprises a power source (e.g. battery) 20, processor 30, a sound module 40; and a sensor module 50. The fishing system 100 is configured to couple to a computer 10 and a display unit 60. The computer 10 may connect to the internet or data in a cloud 80 to enable connection to additional external data inputs 70. This system may beprovided or integrated in a vessel’s (which may be referred to as a boat without loss of generality) hardware.

[0251] Figure 3B shows a block diagram of a second example of the fishing system 100 configured as a permanently installed system. This may be a hidden aftermarket module or an OEM factory installation for example. In this configuration, data processing unit 31 is designed as a standalone processor(s) that can be concealed within the vessel's infrastructure (e.g., beneath a console or inside an electronics locker). Processor 31 may still have video processing 301, processing 302 and edge training 303 portions or processors. Because it is not designed to be portable or modular, system 100 can utilize a larger form factor housing significantly more powerful edge-computing 303 hardware. In some cases, the system 100 relies on discrete, permanently mounted sensor modules 50 and / or sensors 500 such as cameras distributed around the vessel. Alternatively, the system used physically integrated sensors 500. Power is supplied directly and continuously by the main boat battery 20. The system may use the vessel's onboard networking hardware (e.g., NMEA 2000 or marine Ethernet) to interface directly with native marine multi -function displays and cloud 80 infrastructures, acting as a behind-the-scenes engine.

[0252] Figure 3C, shows the fishing system 100 may be coupled with external devices to provide some of the features of the system. For example, a smartphone 32 and charger 34 may be used to provide at least part of the data processing, and / or one or more sensors 500, such as an external camera 42 or GPS sensor 43may provide sound and / or visual data input.

[0253] Figure 3C shows a block diagram illustrating a third embodiment of the fishing system 100 configured as a lightweight, distributed architecture. This may leverage a user's existing external devices 32. In one example, a mobile device (such as a personal electronic device, such as a smartphone) provides the primary computing power, display interface, and network connectivity. An external camera 42 (such as an external action camera), or the smartphone camera (or other mobile device camera), may serve as the primary visual data input. The external camera may be mounted on the short and / or on a pole to provide a suitable field of view. The input may be continuously transmitted to the mobile device 32. The mobile device may execute the models locally to detect events, manage the video buffer, and / or assemble the final media outputs. This may provide a highly portable deployment option. In some cases, the mobile device 32 is configured to connect to a remote server 45 (e.g. through cloud 80) to offload some of the processing. This may depend on the data transfer rates available.The fishing system 100 may provide an automated data capture experience across the complete fishing experience. As seen in Figure 1 systems for fishing may be focused on aiding situational awareness for the on-water experience. These may display environmental or condition data upon which the user may use to make decisions.

[0254] In some examples, fishing system 100 unobtrusively captures performance data during the on-water experience. In some cases, this performance data may be integrated with post-fishing activities to support competitive, educational, social, emotional, and motivational requirements. In some examples, system 100 enables users to enjoy a frictionless transition between on-water performance and off-water insights, creating a comprehensive journey of self-improvement and community engagement. This approach allows both real-time insights and post-trip analysis, ensuring that users can focus entirely on fishing without being burdened by data logging. Historical data may be used in the preexperience phase to assist in planning and preparation to support success with future on-water experiences.

[0255] Figure 3D shows a block diagram of an example fishing system 100 configured as an asynchronous, post-trip media processing architecture for historical or third-party media (Bring Your Own Video). In this example, the system 100 ingests pre-recorded raw media from external sources 190 - such as legacy SD cards, third party cameras, or historical video archives. The raw media may be supplied via a user device 32 or web portal. To accommodate varying hardware capabilities, the system may employ a dual-pathway processing architecture. In a local-edge pathway 111, the user device synchronizes a lightweight version of the model to execute the event detection, automated trimming, and highlight assembly locally on the device. In a cloud 80-processing pathway, the raw media is uploaded to a dedicated server 45 with one or more processors 112 configured to execute one or more models. This may include the event detection and / or video processing models. In both pathways, the system 100 outputs the automated trip report and / or highlight reels 113 (a video summary) without requiring the processor or model to have been present during the initial recording of the footage.

[0256] As seen in Figure 2, the fishing system 100 may offer a user a wholistic unobtrusive experience across all phases of the fishing experience: pre, during, post and between. The interaction or access with the system 100 may vary across the different phases of the wholistic experience for a user. When the user is off the water (phases post, pre and between), the user may interact via a computer 10 or other connected device such as asmartphone 32 to view store data and insights. For example, historical records may be used in the “pre” phase to plan for a future fishing experience, share results with others, identify key learnings from successful and unsuccessful past experiences. The information may be provided in a web-based application or downloadable software app. In some cases, the system may use outputs saved during fishing rather than operating in real time.

[0257] During the on-water (during phase) of the experience interaction may expand to performance information being monitored using sensor 42, 500, 501, recorded and / or processed by processor 30 and optionally displayed on components such as displays 60 of the system 100 or other connected devices 10, 32. Whilst the primary goal of the system 100 may be to record and support the experience on and off the water the system 100 may provide several ways of interacting with the system 100 without interrupting a user experience.

[0258] For example, using voice, gestures and / or sound (instead of having to directly / physically interact with a computer or display screen). For example, specific gestures may be detected and assigned to indicate a particular action, such as a pointing downwards may indicate identification of a waypoint or waving may automate some other type of action, such as turn on video etc. These actions may be interpreted as events and processed with fishing events and / or sent to the video processing processor..

[0259] The fishing system 100 may facilitate configuration of user-specific gestures, sounds and / or voice commands such as clapping, or use of specific words to enhance the usability for a user. In some forms voice commands may be used to start and / or stop the video recording, to take photos of caught fish, measure the fish or other such activities. These voice command activities may be combined with detected events. Furthermore, the system 100 may be configured to continuously analyse audio from audio sensors 40 built-in microphones or wireless wearable microphones for mechanical, environmental, and vocal triggers.

[0260] Mechanical sounds, such as a high-speed clicking or screaming reel drag, or environmental sounds like heavy splashing, may act as events (the events may then act as instantaneous triggers to lock a pre-emptive video buffer). The system may use Natural Language Processing (NLP) to interpret angler dialogue in real-time. Specific spoken phrases (e.g., 'Big hook up!' or 'Get the net!') may automatically increase a quality characteristic of an associated video, or portion thereof. Conversely, phrases indicating a negative outcome (e.g., 'Dropped him' or 'Spit the hook') may automatically decrease the quality characteristic of an associated video, or portion thereof. In some cases, the audio may also be used to assist detection of events, allowing system to tag the video metadata as, for example a 'MissedStrike' or 'Lost Fish'. This may be in combination with a model based on the sensed output from the visual sensor.

[0261] In some cases, the fishing system 100 is configured to determine fishing events.

[0262] Fishing events may be determined by detecting motion and / or objects. The processing may for determining events may use vision-based Al systems. In some cases, the computer vision models analyse the physical behaviour (e.g., position, movement) of the fishing rod (or other fishing gear) across multiple discrete tracking zones (regions of the at least one fishing rod). This physical behaviour is used to determine events (e.g., strikes, casts, misses). Because the determination is based on the fishing rod (and / or fishing gear) it does not, for example, require a user (angler) to be actively holding the gear or even visible in the frame.

[0263] Possible regions (which may be referred to as discrete tracking zones) of a fishing rod include:

[0264] (i) Rod Tip Zone: This may be used to monitor rod tip twitch. In some cases analysing high-frequency vibrations and / or twitches at the rod tip allows detection of subtle bites;

[0265] (ii) Rod Shaft Zone: This may be used to monitor deflection: analysing macrocurvature of the rod shaft to detect heavy strikes and estimate drag tension. In some cases, the rod shaft zone refers to a mid-shaft zone portion of the rod. This may separate effects at the rod tip and / or butt of the rod. The rod shaft zone may identify curvature (including macro-curvature) and deflection under load;

[0266] (iii) Line Zone: this may monitor line Movement and tension. In some case, visually tracking the slackness and / or tautness of the fishing line between the rod tip and the water surface. This may be configured to detect jumps, twitches, or lateral movement before tension reaches the ro. The line zone may cover the visible portion of the line between the rod and water surface;

[0267] (iv) A butt zone: This may be used to determine the usefs movement of the rod, or underlying rod movement, and

[0268] (v) A reel Zone: this may be used to monitor reel handle rotation. In some cases tracking the rotational speed and movement of the reel handle.

[0269] Further details of these zones are discussed in view of Figure 10.

[0270] The particular set, number, and layout of regions analysed need not be fixed, and may be adapted per camera placement, lens configuration, vessel layout, or fishing style whileremaining within the scope of the system 100. In some cases, the regions are inferred by their shadow and / or silhouette. For example, determining a region, rod and / or movement or position within a region should be understood to include inferring the region and / or movement or position based on a silhouette or mask of the regions and / or rod.

[0271] In some examples, the system 100 detects events via these dynamics of the fishing rod (or optionally other fishing gear) alone. In some cases, supporting information is provided by other sensors after event detection. For example, the system 100 may observing a slack-line bite or a strike on a rod mounted in a rod holder during trolling. This automated detection based on dynamics may ensure events are detected even when an angler is not actively holding the gear or present in the frame. The event detection may trigger following actions, such as automated buffer triggering.

[0272] The system 100 may be configured to monitor multiple fishing rods. In such multi -rod scenarios, the system 100 is configured to spatially map the vessel environment to index and monitor multiple unattended rods. The system may be configured to distinguish between active and inactive rods. When an event is detected, the processor may automatically attributes the strike and the resulting video trigger to the specific rod or physical location (e.g., port side outrigger, starboard stern), generating precise contextual metadata before a user (angler) even enters the frame. In some cases, the rod and / or location are associated with a user allowing the system to associate the event with the user.

[0273] Although the system may be implemented as a trained model, all or portions of the system may be implemented analytically. For example, the system 100 may derive feature signals that capture the shape and motion of the regions over time. Feature signals may include, for example, estimates of rod curvature, rates of change in curvature, persistence of bend patterns, and line slack or tension states. Feature signals may be supplied to classifiers and / or rule-based logic to infer interaction states. Interactions states may include no load, transient impulse, sustained load, and landed or lost fish. In one example the processor can generate these feature signals by executing a kinematic tracking algorithm that maps a geometric spline or polynomial curve along the detected length of, for example, the rod shaft and line in each video frame. By comparing the calculated curvature vectors and spatial coordinates of this spline across sequential frames, the system 100 may generate a time-series deformation signal that is computationally compared against predefined mathematical profiles representing specific fishing events.In other embodiments, the system 100 may employ object detection or tracking networks, and / or end to end learned models that operate directly on video and audio streams to output event labels and timestamps. These may not require explicit exposure of curvature as an intermediate signal. In further embodiments, multi modal Al models, including vision language or other large language models, may be used to consume clips or sequences and, in response to prompts, return structured descriptions of casts, strikes, missed strikes, landed fish and other events, which can be used directly for event logging, buffer control, and generation of narrative trip reports.

[0274] Actions, such as position and / or movement of the rod can be indicative of events. For example, strong event signals are provided by:

[0275] • Casting motions: rapid rod swing with line release, followed by line straightening and lure or bait entry into the water.

[0276] • Non-casting bait deployments: controlled line payout with minimal rod swing, such as lowering live bait over the side, dropping weighted rigs vertically, or feeding line while drifting or trolling.

[0277] • Fight / wind-in phases: characteristic alternation between

[0278] (i) high rod curvature with little or no reel rotation and rapid line payout (fish running against drag), and

[0279] (ii) reduced curvature with active reel handle rotation and line retrieval (angler gaining line), sometimes accompanied by “pump and wind” patterns where the angler lifts the rod to increase curvature then winds down as the rod is lowered. Across the fight, the line remains predominantly taut and the rod rarely returns fully to a no-load state until the end of the event. • Landed / boated fish: transition from sustained fight dynamics to rod straightening, reduction or cessation of reel rotation, and localized line slack near the boat or bank. This is often followed by the fish being lifted clear of the water into the frame (for example, hand-landed, netted, or swung aboard), brief gear stillness while the fish is displayed or measured, and then either the fish exiting the frame toward a storage location or being returned to the water.

[0280] • Dropped / lost fish: abrupt transition from an active fight state (high curvature and / or drag noise) to near-zero rod curvature with line suddenly going slack, optionally followed by brief rod lifts or reel turns without renewed tension and angler body-language or vocal cues indicating a loss.Other events may provide support for event classification, or provide lower certainly of an event. For example:

[0281] • Subtle bites: high-frequency rod tip vibration or sudden line twitches and

[0282] slack-to-tight transitions without immediate sustained bend.

[0283] • Ongoing bait presentation / non-event deployment: slow line movement away from the boat with a largely static rod, indicating bait being walked, drifted, or trolled through the water without a clear bite or fight.

[0284] • Missed strikes: brief rod tip or line deflection followed by rapid return to a no-load state without a subsequent fight phase.

[0285] Some actions allow, for example, disambiguation of edge cases, for example:

[0286] • Reel drag audio patterns (for example, sustained high-speed clicking during runs vs intermittent clicks during controlled pressure).

[0287] • Water splash sounds and spray near the apparent line-water intersection.

[0288] • Visual appearance of the fish at or above the gunwale, including being held up for display or measurement.

[0289] • Angler vocal triggers detected via natural language processing (for example, “fish on”, “he’s running”, “get the net”, “dropped him”, “spit the hook”, “keep this one”, “let him go”, “back he goes”).

[0290] In some cases, the end of an event (or event sequence) is inferred from a combination of short-term gear stillness, the fish being visible above water, and / or subsequent actions. Subsequent actions may include: placing the fish into a bin, livewell, or storage area (classified as kept) or returning the fish to the water (classified as released). These outcome labels may be stored in the trip record and / or associated with each completed event.

[0291] In some cases, the training data and / or model structure is configured to prioritise specific mechanical gear interactions such as outlined above. The model, or post-processing may also identify event sequences (which may be referred to as temporal transitions). For example, no-load —> deployment —> bite —> fight with runs and pump-and-wind cycles —> land —> keep or release or drop). Event sequences may provide more robust detection as the system is less dependent on individual events. By concentrating on rod dynamics, the system may achieve robust detection of events (casts, bait deployments, fights, landed fish, lost fish, and outcome states) across diverse fishing styles while remaining compatible with low-power edge compute.In some cases, the system 100 is configured to additionally analyse angler body language. This may be enabled when an angler is detected using an active rod. The system 100 may be configured to recognize postures and / or movements associated with events.

[0292] Example events include: casting, setting a hook, fighting, and landing a fish.

[0293] In some examples computer vision serves as the primary detection method. In some examples fishing system 100 uses non- contact sensors, such as range sensors and / or radar and / or Lidar sensors. An example sensor is mmWave radar. These sensors may supplement the tracking of these movements. These sensors may replace the computer vision sensor. This may require new training data and / or a review of the detail available from non-vision sensors.

[0294] In some cases, the system 100 is configured to determine one or more characteristics of the fishing gear and / or fish and / or vessel. For example, the system may determine fish species, estimate fish size, determine lure types, and detect environmental context. In some cases, these determinations are determined by classification models based on the sensors. For example, the system 100 may use a trained classifier to determine the type and / or size of fish. Such classification systems are known and / or may be implemented by an LLM or VLM.

[0295] The fishing system 100 may be configured to process sensor outputs (sensor data) in real-time. This may be achieved through design of the processor steps reducing computational effort required. In some examples the processing is performed locally. Local processing may be via edge-based computing. This may provide instant analytics, low latency, and offline capability.

[0296] In some cases, the present disclosure described a system configured to process video based on determined events. This system may be linked to the fishing system 100. The video processing system may optimize storage and processing using a multi-path data architecture. For example, a main video data stream may feed directly into a continuous rolling ring buffer (e.g., in passive memory). In a rolling buffer expired frames are automatically overwritten unless a trigger signal (i.e. an event notification) is received by the system (which may be a secondary Al compute layer). If an event notification is received the system may fetch the buffered data, may determine clip boundaries, may identify metrics (such as fish species and size, fight statistics, and trip summary metrics). In some cases, the system executes an optional continuous full-trip save, routing the main video data stream directly to an archival local storage drive (e.g., an onboard SD card or NAS).

[0297] The video processing system may retain this full, continuous trip video. The system may use event data as semantic metadata to tag specific events and narrative B-roll within themaster video file for external editing. Alternatively, or also, the user may configure the system may save only a portion of the trip video. For example, only the trimmed high-res clips and extracted high-res still photos may be saved. The system may actively discard all non-action 'dead water' to conserve storage capacity.

[0298] For off-boarding video, the system may support both wireless and wired data transfer to various external endpoints. A user may select specific wireless synchronization profiles to transmit data to a mobile device, personal computer, or cloud backend, ranging from fullresolution raw video, to compressed proxy clips, down to a 'stills-only' or 'data-only' mode. Furthermore, the hardware module may include an internal integrated battery to facilitate offline, off-vessel data management. This allows a user to utilize a direct wired data connection (e.g., via a cable) to seamlessly transfer heavy media files directly to a personal computer or personal external storage drive.

[0299] The fishing system is preferably configured to automatically detect and log key events such as rod castings, fish bites, reeling activities (fights), fish landings (catches) and release or keep. Each key event may be detected and logged with detailed statistics including, but not limited to, strike-to-cast ratio, fight duration, and fish size estimation.

[0300] In some forms the fishing system 100 is configured to generate trip records to provide trip insights. The trip records preferably aggregate a range of parameters such as environmental conditions, fishing performance metrics, and video content. The system 100 may use an Al -powered semantic video editing engine to construct automated narrative arcs. The system 10 may be configured to identify and stitch together portions of video from an event sequence. In some cases, the most exciting moments — the setup, the tension / fight, and the resolution / landing — are highlighted in a cohesive, shareable highlight video The less exciting moments may be deleted.

[0301] In some cases, the system 100 is configured to synchronize timestamps across multiple disparate sensor (camera) feeds and / or and to select camera viewpoints during detected events. These feeds may include built-in boat-mounted cameras, third-party action cameras (e.g., GoPro, Insta360), wireless sub-surface lure cameras, and angler- worn POV cameras), and aerial camera feeds from unmanned aerial vehicles (UAVs) or drones. POV cameras may include camera-enabled smart glasses and head-mounted displays. The system 100 may be configured to evaluate one or more characteristics (for example, action visibility, subject proximity, and composition quality) to select from and / or combine camera angles(e.g., switching from a sub-surface view of a fish striking the lure to an above-water view from the angler's smart glasses)

[0302] In some cases, the system 100 identifies key moments (which may show peak-action milliseconds -such as the exact moment a fish breaches the surface or is lifted by the angler). Where key moments are identified still the system may extract (optionally high-resolution) still images. This may be from the video and / or external cameras). The still images may populate curated photo galleries within the trip record.

[0303] In some cases, trip insights are logged for access by prediction model (which may be implemented by artificial intelligence (Al) components) of the system 100. The prediction model may be configured to provide insights with actionable recommendations based on historical data and real-time conditions determined by system 100. This may help users improve their fishing techniques.

[0304] In some cases, the prediction model analyses user technique and gear performance. The prediction model may explicitly track retrieval speed, hookset timing, specific bait deployment and casting techniques (such as side-arm, backhand, roll-casting, pitching, flipping, or overhead casts), and specific lure presentation patterns (such as jigging frequency, retrieve-and-pause cadence, topwater walking-the-dog motions, or trolling erraticism). The system 100 may correlate these technical mechanics with specific gear selections (e.g., lure types, rod actions, line types) and real-time environmental variables (e.g., time of day, weather, water temperature, tide states). By aggregating this data over time, the system 100 may generate highly personalized, predictive recommendations, advising the user on optimal gear setups and techniques based on the conditions they are currently facing.

[0305] In some cases, the system 100 models the broader trip context over time to identify non-fishing events and / or to add to fishing events. By combining environmental signals, scene understanding, and sensor data (such as boat speed exceeding a certain threshold, GPS movement, or time of day), the system can recognize story events. Story events are nonfishing events, such as boat departures, high-speed transits between fishing spots, or scenic moments like a sunrise, visible bird activity, or the emotional reactions of the anglers (e.g., capturing the 'face of a bass fisher' after a loss or catch). The system 100 may trigger videos of these moments to save as additional clips of high resolution images. They may be considered narrative 'B-roll', enriching the final Trip Report beyond pure fishing events to provide a complete, broadcast-ready account of the day on the water. This automated triggerand buffering feature may also be beneficial for use in competitions such as Major League Fishing in the USA, providing real-time, validated footage of catches.

[0306] Figure 8 shows a boat 200 in a non-fishing action. The boat may be driving between fishing locations, for example. One or more sensors on boat 200 may be configured to trigger an event. In one example sensor 500 is an image sensor and is configured to detect image features, such as a sunset 800, wave (or a wave of certain amplitude), other boat. In some cases, boat characteristics 820 like speed and or movement may trigger an event. These may be determined by a connection to the boat system. In some cases, and audio sensor may trigger an event based on sounds. The audio sensor may be part of the sensor 500, or a separate sensor.

[0307] Figure 9 shows how the system 100 can determine an event based on rod 600 movement alone. The user is not present in the field of view of sensor 500. However, rod 600 has movement, or a curvature position, which indicates a possible bite. The system may therefore trigger an event more rapidly than if a user was required to be present. This may ensure that video is captured of the entire event and any subsequent events. A portion of line 601 of rod 600 may be occluded from sensor 500 by the stern of boat 200. Therefore, the system 100 may have to determine the event based on rod 900 dynamics alone, or rod 900 dynamics with the partial view of line 901.

[0308] In addition to real-time, edge-based operation, the system 100 may support alternative non-buffer and post-hoc processing modes. In these embodiments, the system 100 acts as a standalone software engine (e.g., a desktop application, mobile app, or cloud portal). The software engine may be designed to ingest pre-recorded, historical, or third-party 'Bring Your Own Video' (BYOV) footage uploaded by a user from external cameras or storage devices. Once ingested, the system 100 may, execute locally on a user's device or remotely via a cloud-based compute architecture, apply the described analysis (e.g., rod, reel, and line dynamics) as well as environmental and trip-context analysis (e.g., detecting scenic moments, boat transits, or weather changes) to this historical footage to retrospectively detect events. The system 100 may then form a highlight video by, for example, trimming the non-action 'dead water,' tags the metadata, and stitch together narrative highlight reels and trip summaries, generate the same comprehensive Trip Report without requiring the use of proprietary on-vessel hardware during the initial capture.

[0309] In some embodiments, the aggregated historical records, trip records, raw or compressed video, extracted still images, and logged performance statistics are furtherutilized as a supervised and semi -supervised training corpus for continual improvement of the system, or the models used by the system. For example, including the event-detection, recommendation, and models. User-confirmed events, corrected misclassifications, and tournament-verified data may be treated as labelled examples, while unlabelled or weakly labelled video segments may be incorporated via semi-supervised learning. Model updates may be computed centrally on an aggregated corpus or via federated learning. In federated learning, the processor or coupled edge devices (such as a smartphone, marine electronics, or smart glasses) perform local training on their own data and transmit only model gradients or anonymized parameter updates to the cloud, without uploading the underlying raw media. A coordinating cloud service aggregates these updates to produce improved models and redistributes the updated models back to the fishing system, enabling the system to learn from fleet-wide fishing activity while preserving user privacy and significantly reducing rawdata transfer requirements.

[0310] In various embodiments, the system models may be trained, fine-tuned, or updated using any combination of supervised learning on labelled fishing events, unsupervised or self-supervised learning on large corpora of unlabelled trip data, semi-supervised learning that mixes a small labelled subset with a larger unlabelled subset, reinforcement or feedback-driven learning that treats user interactions (e.g., keeping, sharing, or deleting an automatically edited video highlight) or tournament outcomes as rewards, and continual or online learning techniques that incrementally update model parameters as new trip records are generated, without requiring full retraining from scratch.

[0311] In some forms the fishing system may synchronise seamlessly with other electronic devices such as smartphones, marine electronics, and cloud-based analytics, allowing users to review, control, compare, and share performance insights across devices. In some forms the fishing system may facilitate live updates of information and / or videos with others.

[0312] In some cases, users of the fishing system 100 have complete control over their data, with the ability to choose between private or public sharing of trip records and achievement milestones.

[0313] Figure 4 shows an embodiment of a sensor module 50 according to the invention. The sensor module 50 may comprise one or more sensors 500. The sensors 500 may be configured to record video. The sensor 500 outputs may be used to detect fishing events. The sensors 500 may provide full 360-degree visual capture. The sensor module 50 may also comprise additional sensors 501. Additional sensors 501 may be configured to measure otherdata points such as temperature, pressure, light etc. The sensor module 50 may comprise an attachment interface 52 configured to attach to a surface. Two example attachment interfaces 52 are shown. The attachment interface 52 may vary depending on the surface or desired use. In some forms the attachment interface 52 is integrated with the sensor module 50 and it other forms the attachment interface 52 is configured to be coupled to the sensor module 50. The attachment interface 52 may allow direct attachment to an accessory mounting system 54 or to an extender component 58 to provide increased placement flexibility of the sensor module 50 to allow improved sensor coverage. The attachment interface 52 may directly or via the extender 58 attach to a base / clamp 54 configured to attach the sensor module 50to the vessel / boat / vehicle or environment it is being used in. The sensor module 50 may be configured to deliver captured data to the computer 10 and / or display 60. The captured data may be delivered wirelessly or using a wire 56. The wires 56 may also provide power to the sensor module 50. Power may be provided by an integrated battery or by connection to an external battery source (not shown).

[0314] To support the universal deployment architecture across varying environments, the attachment interface 52 may be secured to the vessel (boat), angler, or environment using a variety of interchangeable mechanisms. In various embodiments, the mounting hardware may include, but is not limited to, rail mounts (e.g., standard marine T-tracks), rod-holder inserts, suction-cup mounts, magnetic bases, gimbal brackets, permanent flush-mount bezels, threaded tripod mounts, or adhesive plates. Furthermore, the attachment interface 52 may comprise a smart quick-release base incorporating integrated power and data connections (e.g., pogo pins, direct contact pads, or inductive coupling). This allows a user to seamlessly snap the portable sensor module 50 into a pre-wired dock on the boat to instantly receive continuous main-boat power and wired network connectivity, while remaining easily removable at the end of the session for security and at-home data offloading. Additionally, the mounting system may include vibration-dampening materials or mechanical gimbals to stabilize the vision-based sensors during high-speed transit and rough marine conditions, improving the accuracy of the Al event detection.

[0315] In some cases, the system 100 supports an unmanned aerial vehicle (UAV) or drone integration. In this case the sensors and / or the processor are mounted on or integrated within a drone. The drone may be configured to track the vessel or angler. This provides an aerial perspective to continuously monitor the fishing environment, detect events, and / or communicate wirelessly with the system.In some forms, a sound module 40 may be integrated with the sensor module 50 to include a speaker and a microphone.

[0316] Figure 5 shows an example of a fishing system 100 on a boat 200. The system 100 may comprise more than one sensor 500 (which may be packaged as a sensor module 50) at different locations on the boat 200 to capture events from different angles or occurring at the different locations. In some cases, the sensor module 50 is a sensor 500. The sensors 500 may be coupled to one or more displays 60 and a processor 30. The coupling (dashed lines) may be wired or wireless. In some forms a network hub 82 may couple the processor 30 with one or more displays 60 and / or sensors 500 forming a network. As shown sensors 500 may be positioned at the bow and / or mid deck. Inactive rods 600 are shown on the fore deck of the boat 200. The sensors 500 may be configured so that these rods are in the sensor field of view during use. This provides good coverage in the illustrated boat 200. Other sensor 500 locations are possible.

[0317] Figure 6 illustrates an example of a fishing system 100 in action on a boat 200 when a fish or catch 300 is being caught on a rod 600 held by user 650 (who may be referred to as an angler). In this example two sensors 700, 701 are shown to record images (in this case of the fish or catch 300). In some forms the catch event may be captured by one or both of the sensors 700, 701. The catch event may be recorded using one or more sensors 700, 701. The sensors 700, 701 may be of different types, including a radar and video. The processor 30 may determine a ‘catch’ event based on either and / or both sensors 700, 701. The video processing system and / or model may determine which video (or both videos) should be included in a highlights package. In this case the rear sensor 700 may be partially occluded by the angler 650 and / or bow of the boat so the bow sensor 701 may provide a preferably view of the catch 300 breaking the water surface 660.

[0318] In some cases, the fishing event is detected as soon as any strike occurs, such as via movement of the rod 600 and / or movement of the user 650. As described such an event detection may trigger the system 100 to record, or pull from a buffer, recording of the event. This ensures that the event is fully captured not the action after the event. In some cases, the fishing system 100 records over an event sequence. An event sequence may include video / sensor output of an entire fishing event from strike to landing the fish. One or more of the sensors 500, 700, 701 or an alternative output device such as a speaker, may be configured to generate sounds or notifications of commenced or detected events such as “catch” or “video started”.Figure 7 is another example of a fishing system 100 in use on a boat 200. As in the above examples the fishing system 100 may comprise a plurality of sensors 700, 701. The sensors 700, 701 may be provided at different locations on the boat 200. As shown the one sensor 700 is configured to detect the type and measuring the size of the catch 300 or fish once caught, (it may also be used to detect events). A speaker in one of the sensor modules 700, 701 may be configured to notify the user of the identified information. Additionally, or alternatively, the identified information may be recorded and saved in a historical record or associated with the event or event sequence for subsequent review and analysis. This measurement and / or classification process of a caught fish may take place after an event. An event (such as a landed fish) may cause the system 100 to instruct a user 650 to perform the measurement and / or classification process.

[0319] In some forms, the fishing system 100 is capable of distinguishing between different users 650 and / or capturing individualized performance data for each user 650. To achieve this in multi -angler environments (e.g., group fishing trips or competitive tournaments), the system may use computer-vision analysis to track and identify individuals 650. In other cases, the system 100 receives associations between users 650 and fishing rods 600 and / or regions of a vessel 200. Visual identification may include facial recognition, clothing colour and pattern tracking (Person Re-Identification), biometric posture profiling, location, and gear analysis. This visual analysis may optionally be fused with short-range wireless beacon technologies, such as Ultra-Wideband (UWB) positioning tags, Bluetooth Low Energy (BLE) beacons, or other Radio Frequency (RF) tags worn by or attached to the anglers. By correlating spatial position data with visual and biometric event detection, the system can robustly attribute specific events (e.g., casts, strikes, and landed fish) to the correct angler 650 even within shared, multi-person camera views. This attribution facilitates collaborative, team-based trip records while also generating tamper-evident, cryptographically secure chain-of-custody evidence packages for tournament adjudication.

[0320] In some cases, the fishing system 100 may be designed to integrate with existing marine electronics, such as Marine Display Systems (MFDs) and third-party devices, including smart reels, sonar, and GPS, thereby enhancing the overall fishing experience. The fishing system 100 may further support modular add-ons and external accessory sensors to provide deeper insights and verified analytics.In some cases, the system is configured to pair wirelessly with third-party accessories such as Bluetooth-connected digital fish scales, electronic measuring boards, and electronic tagging devices. Other weighing method may be used.

[0321] In an example case, when a stabilized weight or length measurement is detected by an external accessory (instead of, or as well as, sensor 700 as shown in Figure 7), the system 100 is configured to associate this data with one or more events and / or event sequences. This may be through time synchronisation. The system 100 may stores it as structured metadata.

[0322] Additionally, the system may use a model to automatically calculate 'time-out-of-water' conservation metrics by visually tracking the exact duration a landed fish remains out of the water before release. This data may be securely logging this metric for tournament compliance and fish welfare verification. This integrated data may then be used to automatically generate visual data overlays (e.g., displaying the verified weight of the fish on screen) within the curated media of the Trip Report. Furthermore, the system may integrate additional environmental sensors, such as light sensors, barometers, and temperature sensors, to enrich the data, events, features and predictive analytics.

[0323] Furthermore, the fishing system 100 may be configured to automatically compare the size of a caught fish 300 with pre-defined minimum legal-size requirements as established by local regulatory authorities, immediately alerting the user to facilitate regulatory compliance and support conservation efforts. In some forms, a user may be enabled to set their own minimum size requirements that meet or exceed regulatory requirements for specific fish species. The length of the fish 300 may be detected by the fishing system 100 and / or connected systems. Alerts or notifications may be issued indicating compliance with the set requirement.

[0324] In some examples, the fishing system 100 may be built for integration with events such as professional fishing tournaments, providing real-time footage, live streaming, leaderboards, and / or accurate scoring for events like Major League Fishing. Furthermore, the fishing system 100 may provide a seamless experience for both participants and viewers. Other embodiments of the fishing system may incorporate additional or alternative components or sensors including:

[0325] • Manual Tracking input: Users may manually count their casts and fish catches using a traditional counter or a mobile app. This method is straightforward but requires active user involvement and is prone to human error, making it less efficient than an automated solution.• Low Power Sensors: A low power sensor such as a Bluetooth™ sensor may be coupled to a fishing rod or other equipment to detect motion and trigger related to casting or fish catches. While this provides some automation, it is less precise and versatile than the vision-based system and requires pairing with a mobile app or device for data tracking. Also attaching components to fishing rods can alter the balance of the rod so therefore impacting the efficiency.

[0326] • Accelerometers and Motion Sensors: Other types of sensor such as accelerometers and motion sensors may be embedded into or coupled to the fishing rod or worn by a user or other equipment to detect movement patterns associated with casting and fish catches. For example, wearable devices such as a watch type or ring type device or glove worn on the wrist or hand may include embedded sensors to detect movement by the user to determine casting actions and types. This approach offers automation but is likely less accurate than a vision-based system but may struggle with detecting subtle movements. Furthermore, calibration for each user and fishing context may be required but is likely to add complexity.

[0327] • Traditional Radar-Based Systems: A radar system may be used to detect motion or proximity of objects in the environment. While radar works in poor lighting or weather conditions it lacks the detail and resolution of a camera system, making it less accurate for tracking and counting individual actions. Radar may be integrated to provide distance calibration for events such as casting.

[0328] • Lidar-Based Systems: A Lidar system may be used to map the environment and detect movement. Lidar provides accurate distance measurements but struggles in certain outdoor conditions and lacks the detail in object recognition or tracking compared to a vision-based system. It is also more expensive and bulkier.

[0329] • Infrared-Based Vision Systems: Infrared systems detect heat signatures and can track casting actions and detect fish based on temperature differences. However, they have limitations in resolution and detail needed for precise counting of casts and catches. Infrared systems are also more expensive and require specialized processing hardware.

[0330] The disclosed fishing system 100 may provide one or more of the following features or advantages:

[0331] • The fishing system provides a hands-free operation with no manual input required, ensuring a fully immersive, distraction-free fishing experience.• The fishing system may recognize and differentiate between multiple users or anglers, making it ideal for group fishing trips and competitive events such as tournaments. • The fishing system may provide a leaderboard, trip challenges, and personal milestones, offering a fun, competitive experience for fishing enthusiasts. Users may compare performance, engage with others' achievements, and track progress over time.

[0332] • Users may log waypoints, mark catches, and interact with marine electronics using simple voice commands and / or gestures, making it easier to capture key moments while fishing.

[0333] • The fishing system may analyse historical trip data and expert fishing strategies, providing predictive insights and tailored recommendations to improve fishing success, ensuring users know the best tactics for any given conditions.

[0334] • The fishing system may be configured to adjust to varying environments, including freshwater, saltwater, and tournament settings, factoring in environmental variables such as light, weather, and water conditions to ensure optimal performance tracking.

[0335] • The fishing system may be configured to automatically start, stop and / or edit data and videos to compile key moments from fishing sessions, such as the strike, so that users can easily relive their most exciting catches.

[0336] • In some forms the fishing system may include additional sensors such as light sensors, barometers etc. and / or additional cameras to provide broader environmental coverage and more accurate tracking.

[0337] In one example the fishing system 100 is a modular, edge-based system that integrates sensor fusion technology to deliver real-time, hands-free fishing analytics. The system 100 may be designed to work with both standalone hardware and existing marine electronic devices, offering flexible deployment options. In one example the components and manufacturing approach include a sensor fusion module with mmWave radar and / or other sensor to capture precise casting motion, fish strikes, and fight dynamics, a vision-based system including a camera and model to identify fish species, lure types, and environmental conditions, and an edge processor to run onboard models for real-time event detection without requiring cloud connectivity.

[0338] The hardware design and manufacturing considerations may include a waterproof and marine-grade housing to ensure durability in harsh fishing environments (e.g., IP67 / IP68 rating), universal mounting system compatible with boats, kayaks, docks, and potential rodattachments, and integrated power management to support boat-powered, battery-operated, and solar-assisted configurations.

[0339] The software and data processing may involve edge-based processing to reduce latency and allow real-time feedback, mobile and marine electronics integration to sync with smartphones, MFDs (Multifunction Displays), and cloud storage, and automated video processing and trip records or report using the system processor to highlight key fishing moments. The connectivity and user experience may include offline functionality to work without an internet connection, synchronising when connectivity is restored, voice-activated controls to facilitate hands-free interaction with marine electronics and logging, and gamification and social sharing to enable leaderboard tracking, social challenges, and trip record sharing.

[0340] The fishing system 100 may be designed to be non-intrusive, high-performance, and adaptable across different fishing environments. The system may be sold as a standalone device or offered as a software integration for existing devices such as to marine electronics. The disclosed fishing system 100 may be implemented in multiple configurations. To ensure the event-driven media and analytics workflows operate regardless of the physical hardware format, the fishing system 100 may be deployed across a Universal Deployment Architecture comprising at least three primary configurations:

[0341] (i) Portable / Removable Aftermarket (Integrated or Modular): A portable aftermarket system that may be deployed in either an integrated or modular physical architecture. In an integrated configuration, the edge processor, cameras, and sensors are housed together within a single, self-contained waterproof device. In a modular configuration, the system comprises a central compute module that connects to external proprietary cameras or third-party action cameras. To overcome the battery life and continuous-recording limitations inherent to standard action cameras, the central compute module is configured to supply continuous power and receive high-bandwidth data streams from the external cameras via direct wired connections, physical docking interfaces (e.g., magnetic pogo pins, hot-shoes, or proprietary snap-on mounts), or wireless power and data transfer protocols. The device includes an attachment interface to facilitate temporary mounting to a water vessel for a session, and can be removed by the user for direct off-board syncing or charging.

[0342] (ii) Permanent Aftermarket Installation: A modular system permanently wired and / or bolted to a dedicated vessel (e.g., a charter boat). This configuration is optionally integratedwith the vessel's power supply and onboard network, facilitating automatic dockside or harbour syncing (e.g., via Wi-Fi) when network connectivity is restored.

[0343] (iii) Factory-Integrated OEM: A configuration where the cameras, sensors, and computing processors are permanently integrated into the vessel's structure (e.g., hardtop, console, or hull) and / or built directly into the vessel's native marine electronics (e.g., Marine Display Systems or MFDs) at the time of manufacture.

[0344] (iv) Unmanned Aerial Vehicle (UAV) or Drone Integration

[0345] A configuration where the visual sensors and Al compute layer are mounted on or integrated within a UAV or drone tracking the vessel or angler. In this configuration, the drone continuously monitors the fishing environment from an aerial perspective, utilizing the same computer-vision behavioural analysis of the gear dynamics and line movement to detect strikes and trigger buffer retention, optionally communicating wirelessly with the vessel's central compute module to aggregate the media into the unified trip record.

[0346] (v) Mobile Device or Smartphone Integration

[0347] A configuration where a consumer mobile device (such as a smartphone or tablet) serves simultaneously as the visual sensor, the Al compute layer, and the storage medium. In this embodiment, the smartphone is mounted to the vessel, and its native camera continuously captures the media stream while its native processor executes the computer-vision models to detect gear dynamics and fishing events. To mitigate thermal limitations and prevent overheating of the mobile device during continuous video ingestion and Al inference, particularly in direct sunlight on a vessel, the application may dynamically downscale camera resolution or framerate during periods of no activity, offload specific processing tasks to cloud infrastructure, or utilize localized thermal-management hardware cases.

[0348] Figure 10 shows an angler 650 with rod 600 having line 601. A sensor 500 attached by pole 1010 to boat 200 has a field of view 1050 including rod 600. While a single angler 650 and rod 600 are shown it will be understood that multiple rods 600 and / or anglers 650 may be within a sensor’s 500 field of view. The rod 600 is shown with regions indicated. While sensor 500 is shown alternative sensor positions may be used, include an airborne sensor 1090, such as based on a drone. The system 100 is configured and / or trained to monitor regions of the rod 600 to improve event determination and / or reduce the computational requirements for determination of events.Figure 10 shows a rod tip zone 1020; a mid-shaft zone 1021; a butt zone 1022; a line zone 1023; and a reel zone 1024. In some cases, one or more of these regions may be omitted and / or combine with one or more other regions.

[0349] In some cases, the tip zone 1020 comprises a portion of the fishing rod 900 at or about the tip 905 of the fishing rod. (the tip 905 may be referred to as the distal end of the fishing rod 900). The tip zone 1020 portion may comprise at least the top 5%, the top 10% or the top 20% of the shaft (blank) of the fishing rod 900.

[0350] In some cases, the mid shaft zone 1021 extends from the lower boundary of the tip zone 1020 down to around the reel seat 1030 (where the reel 1031 is to be attached). In some cases, the mid-shaft zone 1021 extends from the reel seat 1030 to within approximately 5%, 10% or 20% of the tip 905 of the shaft of the fishing rod 900.

[0351] In some cases, the butt zone 1022 extends from the reel seat 1030 to the proximal end 906 of the fishing rod 900. In some cases, the butt zone 1022 extends adjacent to a handle 1040 of the fishing rod 900.

[0352] In some cases, the line zone 1023 extends along a fishing line 1060 of the fishing rod 900. In some cases, the line zone 1023 extends between the tip 905 of the fishing rod 900 and a lure 1070 or a water surface 1080, or a portion thereof. In some cases, the line zone 1023 extends along the fishing line 1060 from the reel 1031 to the lure 1070 or the water surface 660.

[0353] In some cases, the reel zone 1024 extends over the reel 1031. In some cases, the reel zone 1024 extends over all, or a portion of, the handle 1040 about the reel 1031.

[0354] In some cases, the entire fishing rod 600 forms a region. In some cases, the line zone 1023 is included as the entire fishing rod 960 region. It will be understood the number, size and boundaries of the regions may be varied. In some cases, the boundaries may vary by, for example, 10% or 50%. In some cases, a user region 1025 extends about the user of the rod 600. The models may be trained to determine events based on user position.

[0355] Figure 11 shows a flow diagram showing an example method for determining events. In step 1100 image data, such as video, of one or more fishing rods is received. This may be real-time data in an edge processing application, or earlier recorded data if being processed later. A preprocessing stage 1101 may be used to determine fishing rods in the field of view of the video. This stage 1101 may determine the number and / or location of fishing rods. This stage may determine if fishing rods are active (i.e. in use) or inactive. In some cases, one or more or each rod is associated with a user in step 1102. This may be through user input,location sensors or otherwise. As shown in step 1125, the associated user may be updated during the process. For example, if the images show a different user has moved to the fishing rod. Stage 1103 divides the video into video portions for each active fishing rod. This may simplify processing because the model 1120, 1121, 1122, only has to process a single active rod in each video. In alternative cases, the models 1120, 1121, 1122 are trained to process multiple rods in single videos.

[0356] Further preprocessing steps may be applied to the video. For example, the video streams may be processed to account for boat movement 1104. This may be based on an external sensor, such as an IMU. Step 1105 shows that further sensors outputs and / or context information may be provided to the models 1120, 1121, 1122. For example, audio output and / or boat movement data may be provided to improve event detection. In some cases, user input is also provided.

[0357] The pre-processed video is output to models 1120, 1121, 1122. As discussed, a combined model 1120 may be used, or a plurality of models 1121, 1122 may be used. Further models may also be used. The two-model system shown uses a first model 1121 to determine 1105 fishing rod regions in each video and a second model 1122 to determine 1106 events based on the position and / or movement of the fishing rod within those regions. The combined model 1120 is configured to implement these steps concurrently. The model 1120 may not make the intermediate steps visible, or explicit.

[0358] In step 1106 the models 1120, 1122 output an event determination. The event determination may include the event and one or more characteristics of the event. Example characteristics are timestamps and associated users. The timestamp may be based on the time or time periods of the input video. Additional sensor outputs and / or context information (e.g. from step 1105) may be associated with the event. For example, a high-quality image from a different sensor and / or audio recording about the event.

[0359] Step 1107 shows the additional output of event metrics. Event metrics may be used to track performance and / or make predictions for improved performance, as previously explained herein.

[0360] Figure 14 shows a flow diagram for the training of the models 1121, 1122 of an example, two model 1121, 1122 system 100. Training data may be provided of previous images and videos. The training data will be labelled. Training data 1401 may be labelled with the fishing rod regions. Training data 1402 may be labelled with events. Training data 1403 may be labelled with rod regions and events. The first model 1121 may use trainingdata 1401 and 1403 as both have labelling of fishing rod (or user or other gear) regions. The second model 1122 may use training data 1402 and 1403 as both have labelling of events. In some cases, all training data is fully labelled as training data 1403. Figure 14 shows that the outputs of the first model are fed to the second model to provide the region determination to the second model, simplifying the event determination.

[0361] Figure 15 shows a flow diagram for the training of a single model 1120. In this case, the training data 1401, 1402, 1403 is the same as in Figure 14 but now all the data is fed into the single model 1120. Again, the model may only use training data 1403 if required, but additional training data 1401, 1403 may assist the model.

[0362] Figure 12 shows an example flow chart for the processing of video data based on event notifications in video processing system 1200. By interfacing the event notifications, and in particular event sequence notifications, with video processing more efficient video processing and video creation can be performed. This may allow the video to be processed in real-time and / or reduce computational and / or storage requirements.

[0363] Step 1201 shows the receipt of an event notification. System 100 may have provided the event notification. Optionally the event notification may have a time stamp so as the system 1200 can associate 1202 other data (such as sensor data and / or user inputs 1220) with the event. In some cases, associated data may be provided with the event notification. The system 1200 may read video data from the rolling video buffer 1203. The buffer ensures that video is captured of and surrounding the event. Using the rolling video buffer reduces video storage space requirements. In some cases, the event notification 1201 is provided by an additional sensor. For example, a user input (GUI), an audio sensor and / or a visual sensor (separate to the primary sensor). This may allow events obscured from the primary sensor to be captured by the system and / or for users to input events.

[0364] In step 1203 the system 1200 determines if the received event 1201 is part of a current event sequence. This may be determined through set logic. For example, the logic may be based on the type of event (initial, intermediate or end events), or by comparison to current event sequences 1204. If the system 1200 cannot determine if the event is a continuation (for example if an initial event has been missed) the system 1200 may create a current event sequence to retain the video until further event notifications and / or information is available.

[0365] If the received event 1201 is not part of a current event sequence 1204 the system 1200 will create a new event sequence 1205. The video from the buffer 1203 and any other associated data (including other video and images) can be saved 1206. The system 1200 maycontinue to save video 1206 to the current event until the event sequence ends. This allow video for the whole event sequence to be obtained. However, an event sequence may be long requiring large video storage. Step 1207 initialises a quality characteristic (also referred to as an importance characteristic). The quality characteristic is indicative of the quality of the event to the video summary, or otherwise indicative of more important moments. The system 1200 may use the quality characteristic and / or one or more user defined rules or logic to determine how to save video. For example, if the quality characteristic is low the system may reduce the length and / or quality of video saved. Alternatively, if the quality characteristic is high the system may save full length and quality video. As discussed above quality characteristic may be determined by the type of event and / or external sensor data. Once the event sequence is open and video is being stored at the appropriate level the system 1200 waits 1208 for the next event. A time out value may be stored to ensure the event sequence is closed. The system 1200 may have a plurality of event sequences open at any one time (in some cases one for each fishing rod).

[0366] If the event is a continuation of a current event sequence the system 1200 will add the event to the event sequence in step 1209. The event video will be added to the event sequence in step 1210. The system 1200 may recalculate the quality characteristic continuously and / or based on new event notifications. Step 1211 shows recalculation of the quality characteristic based on the new event. For example, if the event is a caught fish the quality characteristic will increase and high portions of video and / or additional images may be included. In step 1212 the system 1200 determines if the event is a final event. If so, the video (and / or image) outputs may be finalised 1213 based on the quality characteristics and the series of events and / or any user settings. If not the system 1200 awaits the next event notification 1214. The time-out value may be set and / or adjusted based on the new event. In this manner the system 1200 is able to automatically generate video highlights based on event notifications.

[0367] In some cases, system 1200 does not just record everything during an event sequence. Once an event sequence starts it continues to analyse quality characteristics (optionally in real time) and may use logic, such as time-outs and thresholds, to decide how much video to keep around that sequence or discard. In some cases, the system behaviour is

[0368] mode-aware because trolling / bait fishing and active casting behave very differently.

[0369] In an active lure / bass-style casting example the sequences may be short and tightly coupled. For example: cast —> retrieve —> strike / fight / land (or other end event). In such cases the system may use a relatively short rolling window; save from shortly before the initial castthrough to the landing / release (or end event), and / or close the sequence quickly once rod / load activity returns to idle for a short time.

[0370] In trolling / bait-deployment scenarios there can be a long quiet period between deploy and bite. In some cases when lines are deployed, the system 1200 starts an event sequence and saves a short segment showing the deploy, then allows a low-activity “open” state. The low-activity open state may keep a lightweight state for that sequence and continues to monitor rods, boat motion and audio. If no further events are notified beyond a configurable timeout, the system 1200 may stop writing continuous video forthat rod.

[0371] However, the system may keep the deploy metadata and timestamp. If a later strike / fight is detected on that rod, it pulls pre-event footage of the strike / fight from the rolling buffer, and stitches this to the stored deploy segment. This may enable the system 1200 to form one narrative sequence.

[0372] In some cases, the system 1200 may include adaptive time-outs and rules that decide when a sequence is considered finished. For example: a rod returns to idle for N seconds, boat starts moving to a new location, user marks a stop. At these adaptive time outs the system 1200 may finalises any pending event sequences and associated videos. Any future events are treated as new sequences. Based on user settings the system 1200 might not even keep all the clips if they don’t mean the quality threshold. The system may still store metrics on the unwanted clips. Users may not want to keep images / footage of capturing small undersized fish but will more commonly want to know metrics. These actions may let the system 1200 keep bait deployment as part of the story, scale buffer usage to the true length and intensity of the fight and avoid saving hours of uneventful footage.

[0373] Figure 13 shows a block diagram of example internal data processing architecture and event driven rolling video buffer logic utilized by the system 1200. The architecture begins with the continuous ingestion of raw visual and sensor data. This may be from sensors 500, with additional sensors 501 providing further outputs. This main video data stream feeds directly into a video buffer 1250, which is shown as a continuous rolling ring buffer (which may be in passive memory), where expired frames are automatically overwritten to conserve storage. The system 1200 may continuously monitor for an event notification (trigger signal) from a notification system such as system 100. Upon receiving an event notification the system 1200 fetches the relevant pre- and post-event video segments from the buffer 1250. This provides an initial video of the event which captures video from prior to receipt of the vent notification.Simultaneously, the system 1200 may optionally execute a parallel continuous fulltrip save or otherwise route the main video data stream directly to an archival local storage drive to bypass the buffer. This, and / or the continued processing of data from the main video / image data stream, allows the system 1200 to store video over an event sequence. The architecture of Figure 13 may optimize storage and processing. When the event notification is received, the system 1200 can retroactively transfer a portion of the rolling buffer 1250 into persistent storage as a candidate clip.

[0374] The system 1200 may then analyse that segment to determine precise clip in-points and out-points. This may be determined by a computer vision system. Clip points may be determined by the segment characteristics, determined events in the segment and / or additional sensor 501 outputs. Timestamps may be used to synchronise the video with additional sensor 501 outputs. Additional processing of the clip and / or additional sensor 501 outputs may identify other characteristics. For example, fish species and size where applicable, calculate fight statistics, and generate trip-summary metrics. In some embodiments, the system 1200 maintains a continuous rolling video buffer (for example, 30 to 60 seconds in length) across one or more cameras capturing the environment, and when fishing event notifications are received the system does not simply apply a fixed pre- and / or post- event window but instead uses multimodal signals (including rod and line deformation, reel and water acoustics (such as reel drag and splashes), and vessel motion or speed changes) to determine dynamic buffer extents. This provides more control over the saved videos. Extended periods of “dead water” discarded without manual input.

[0375] In some cases, the event notifications (which may be derived from one or more of rod 600 and line 601 deformation) and / or reel and water acoustics and / or vessel motion or speed changes are used to determine dynamic clip in-points and out-points for the video system 1200. This allows the saved segment to begin before the main burst of activity where the build-up is important, and to extend through to the true end of the fight or moment. Different event types may have different retention policies (for example, saving a longer segment for a landed fish than for a missed strike), and extended periods of non-event “dead water” are automatically discarded without user input.

[0376] The system 1200 may use additional sensor information, such as boat speed variation, and acoustic signatures to compute the dynamic clip in-points and out-points for each event, with buffer length automatically scaled to the true length of the fight andFigure 16 shows an example output from the system for a display 60. The output shows a spatial map 1600 with events 1605. The events 1605 are shown with a location. The location is shown along a travel path 1610 of the vessel (boat 200). In some cases, the display is configured to show a neutral boat travel path 1610. In some case the system displays icons / dots / vectors for events 1605 (such as casts, strikes, catches, drops). Events in event sequences may be colour coded or otherwise grouped. In some cases, the display 60 is linked to the video from the system 1200 and / or associated data. For example, tapping an event 1605 icon may open the buffered clip, photos, and metrics. In some cases, a heat map and / or intensity map is used to show events. For example, the display may show cast / fish intensity via dots / vectors, engagement lines, area grids, or a linked graph. In some cases, the display is configured to recolouring the boat travel path 1610 itself to show events and / or event density.

[0377] In some cases, the system displays metrics, or the described spatial cast-density and effort maps. For each detected cast, strike, fight, landed fish, or missed strike, the system may record the corresponding vessel position and, where available, chart or bathymetric data describing nearby geographic or underwater features such as shorelines, weed lines, drop-offs, channels, and submerged structure. Other sensor and / or environmental data may be included on the display.

[0378] In some configurations, the software methodology — including computer-vision event detection, pre-emptive buffering, semantic tagging, automated media editing, and asynchronous cloud offloading — is executed consistently. Furthermore, the system may provide real-time leaderboard updates and live streaming based on automated strike detection and fish measurement. Such automated updates reduce the need for live camera crews by using the system's native event validation, allowing for live broadcast integration with competitive events such as Major League Fishing. These different embodiments ensure that the system functions adaptably across different mounting styles, vessel types, and competitive or recreational environments without departing from the scope of the invention.

[0379] Examples may be implemented by hardware, software, firmware, middleware, microcode, or any combination thereof. Methods may be implemented as computer implemented methods. When implemented in software, firmware, middleware or microcode, the program code or code segments to perform the necessary tasks may be stored in a machine-readable medium such as a storage medium or other storage(s). A processor may perform the necessary tasks. A code segment may represent a procedure, a function, a subprogram, a program, a routine, a subroutine, a module, a software package, a class, or anycombination of instructions, data structures, or program statements. A code segment may be coupled to another code segment or a hardware circuit by passing and / or receiving information, data, arguments, parameters, or memory contents. Information, arguments, parameters, data, etc. may be passed, forwarded, or transmitted via any suitable means including memory sharing, message passing, token passing, network transmission, etc.

[0380] Methods may be stored on storage medium. A storage medium may represent one or more devices for storing data, including read-only memory (ROM), random access memory (RAM), magnetic disk storage mediums, optical storage mediums, flash memory devices and / or other machine readable mediums for storing information. The terms "machine readable medium" and "computer readable medium" include, but are not limited to portable or fixed storage devices, optical storage devices, and / or various other mediums capable of storing, containing or carrying instruction(s) and / or data, including non-transitory mediums.

[0381] The various illustrative logical blocks, processors, modules, circuits, elements, and / or components described in connection with the examples disclosed herein may be implemented or performed with a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic component, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, circuit, and / or state machine. A processor may also be implemented as a combination of computing components, e.g., a plurality of processors, a combination of a DSP and a microprocessor, a number of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration.

[0382] The methods or algorithms described in connection with the examples disclosed herein may be embodied directly in hardware, in a software module executable by a processor, or in a combination of both, in the form of processing unit, programming instructions, or other directions, and may be contained in a single device or distributed across multiple devices. A software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. A storage medium may be coupled to the processor such that the processor can read information from, and writeinformation to, the storage medium. In the alternative, the storage medium may be integral to the processor.

[0383] One or more of the components and functions illustrated the figures may be rearranged and / or combined into a single component or embodied in several components without departing from the present disclosure. Additional elements or components may also be added without departing from the present disclosure. Additionally, the features described herein may be implemented in software, firmware, hardware, and / or any combination thereof.

[0384] In its various aspects, the present disclosure can be embodied in a computer-implemented process, a machine (such as an electronic device, or a general-purpose computer or other device that provides a platform on which computer programs can be executed), processes performed by these machines, or an article of manufacture. Such articles can include a computer program product or digital information product in which a computer readable storage medium containing computer program instructions or computer readable data stored thereon, and processes and machines that create and use these articles of manufacture.

[0385] The invention may also be said broadly to consist in the parts, elements and features referred to or indicated in the specification of the application, individually or collectively, in any or all combinations of two or more of said parts, elements or features.

[0386] The entire disclosures of all applications, patents and publications cited above and below, if any, are herein incorporated by reference.

[0387] Reference to any prior art in this specification is not, and should not be taken as, an acknowledgement or any form of suggestion that that prior art forms part of the common general knowledge in the field of endeavour in any country in the world.

[0388] Where in the foregoing description reference has been made to integers or components having known equivalents thereof, those integers are herein incorporated as if individually set forth.

[0389] It should be noted that various changes and modifications to the presently preferred embodiments described herein will be apparent to those skilled in the art. Such changes and modifications may be made without departing from the spirit and scope of the invention and without diminishing its attendant advantages. It is therefore intended that such changes and modifications be included within the present invention.

Claims

CLAIMS1. A monitoring system for a fishing rod, the monitoring system comprising a processor configured to:receive an output from the one or more sensors;determine one or more regions of at least one fishing rod from the output; determine one or more events based on the output in the one or more regions of the at least one fishing rod.

2. The monitoring system as claimed in Claim 1, wherein analysing the fishing rod comprises determining the position and / or movement of the fishing rod, optionally in the one or more regions.

3. The monitoring system as claimed in claim 2, wherein determining an event is based on the position and / or movement of the fishing rod in each of the one or more regions over a period of time.

4. The monitoring system as claimed in claim 1, wherein the output from the one or more sensors is a video.

5. The monitoring system as claimed in any one of claims 1 to 4, wherein the determination of an event uses the output of at least two of the regions.

6. The monitoring system as claimed in any one of claims 1 to 5, wherein the regions comprise one or more ofa rod tip zone;a mid-shaft zone;a butt zone;a line zone; anda reel zone.

7. The monitoring system as claimed in any one of claims 1 to 6, wherein the processor is configured to determine the one or more regions of the at least one fishing rod using a first model, wherein the first model is trained by training data comprising a plurality of images including fishing rods wherein one or more of the one or more regions have been labelled.

8. The monitoring system as claimed in any one of claim 1 to 7, wherein the processor is configured to apply:a first model configured to determine the one or more regions of the fishing rod, andP1142NZ00a second model configured to determine the one or more events based on the determined the one or more regions of the first model.

9. The monitoring system as claimed in claim 8 or 9, wherein the first model comprises a vision model configured to detect and label the one or more regions of the fishing rod.

10. The monitoring system as claimed in any one of claims 1 to 6, wherein a model is trained to determine the one or more events from the received outputs.

11. The monitoring system as claimed in claim 9, wherein the model is trained on training data comprising a plurality of images labelled with the one or more regions and the one or more events.

12. The monitoring system as claimed in any one of claims 1 to 11, wherein the processor is configured to determine event sequences.

13. The monitoring system as claimed in claim 12, wherein the processor is configured to determine event sequences algorithmically based on the detected events.

14. The monitoring system as claimed in any one of claims 12 to 13, wherein the processor is configured to determine a quality characteristic for an event.

15. The monitoring system as claimed in claim 14, wherein the system is configured to track a quality characteristic of an event sequence.

16. The monitoring system as claimed in claim 15, wherein the system is configured to determine an amount of the output to save based on the quality characteristic.

17. The monitoring system as claimed in any one of claims 1 to 16, wherein the one or more sensors are located in sensor positions comprising one or more ofthe stem of the boat;the bow of the boat;body worn;mounted to the fishing rod;external to the boat.

18. The monitoring system as claimed in any one of claims 1 to 17, configured to monitor a plurality of fishing rods, wherein the processor is configured to:firstly, detect each fishing rod, andsecondly identify regions on each detected rod.

19. The monitoring system as claimed in any one of claims 1 to 18, wherein the events comprise any one or more of: casts, bites, strikes, landings, fights, lost fish, and misses.

20. The monitoring system as claimed in any one of claims 1 to 19, wherein the processor is configured to determine vehicle movement based on the output of the vehicle position sensor and:process the output of the sensor to remove movement of the vehicle, and / or process the output of the sensor to determine movement of the fishing rod independent from the movement of the vehicle, and / ordetermine the relative movement of the fishing rod and the vehicle.

21. The monitoring system as claimed in any one of claims 1 to 20, wherein the processor is configured to apply a high pass filter to concentrate on high-frequency movement in each of the regions.

22. The monitoring system as claimed in any one of claims 1 to 21, wherein the processor is configured to associate one or more sensor outputs with the determined event and / or associate one of more environmental conditions with the determined event.

23. The monitoring system as claimed in any one of claims 1 to 22, wherein the processor is configured to:determine a spatial path from the spatial tracker andassociate a spatial path to the one or more events.

24. The monitoring system as claimed in claim 23, wherein the processor is configured to determine, based on the spatial path of a plurality of events, an event map.

25. The monitoring system as claimed in any one of claims 1 to 24, wherein the processor is configured to produce a video summary based, at least in part on the detected event.

26. The monitoring system as claimed in any one of claims 1 to 25 comprising the one or more sensors, the one or more sensors configured to monitor an area including the at least one fishing rod.

27. A computer implemented method for monitoring a fishing rod, the method comprising the steps of:receiving an output from one or more sensors;determining one or more regions of at least one fishing rod from the output; anddetermining one or more events based on the output in the one or more regions of the at least one fishing rod.

28. A system for producing a video summary of a fishing activity, the system comprising:a video storage buffer configured to temporarily store video from one or more cameras; anda processor configured to receive an event notification and save at least a portion of the temporarily stored video.

29. The system as claimed in claim 28 wherein the processor is configured to receive notifications from an event sequence, wherein a first event in the event sequence is configured to save a portion of the temporarily stored video and the processor is configured to continue storing video until a final event notification is received.

30. The system as claimed in any one of claims 28 or 29 wherein the processor is configured to receive a quality characteristic associated with the event and adjust a characteristic of the portion of the stored video based on a quality characteristic.

31. The system as claimed in any one of claims 28 to 30 wherein the processor is configured to save, with the temporarily stored video, future video from the one or more cameras, optionally until the notification of an end event.

32. The system as claimed in any one of claims 28 to 31 wherein the processor is configured to determine a portion or characteristic of the temporarily stored video and / or future video to save, based at least in part, on the event notifications.

33. The system as claimed in any one of claims 28 to 32, wherein the event notification is received from the monitoring system as claimed in any one of claims 1 to 26.

34. A computer implemented method for producing a video summary of a fishing activity, the method comprising:storing received video from one or more cameras in a video storage buffer; receiving an event notification; andsave at least a portion of the temporarily stored video.

Images