Animal collection and identification method of distributed camera and distributed camera system

Abstract

The invention discloses an animal collection and recognition method of a distributed camera and a distributed camera system, and relates to the technical field of camera recognition. The method comprises the following steps: each camera detects the infrared radiation intensity change of a respective target area through a PIR sensor; and when any PIR sensor detects that the change of the infrared radiation intensity of the corresponding target area exceeds a preset threshold value, the corresponding camera acquires an image of the target area through the image acquisition module, outputs corresponding image data, and caches the image data in the local storage module of the corresponding node. Each camera sends image data to the gateway through the wireless communication module according to a preset time interval. And the gateway receives the image data of each camera, inputs the image data of each camera into a pre-trained animal identification model, and generates activity tracks and behavior statistical data of different types of animals in combination with space-time correlation analysis. Therefore, the problems of data dispersion of the cameras and low manual data processing efficiency are solved.

Description

Technical Field

[0001] This application relates to the field of camera recognition technology, and in particular to a method for animal collection and recognition using distributed cameras and a distributed camera system. Background Technology

[0002] In applications such as wildlife monitoring, ecological research, and nature reserve management, outdoor tracking cameras (also known as infrared-triggered cameras) are widely used as non-invasive, low-cost, and long-term deployable monitoring tools to record wildlife activity, population distribution, and habitat utilization. These cameras are typically based on passive infrared sensing technology, automatically triggering image or video capture when a warm-blooded animal is detected passing by, and storing the data on a local memory card.

[0003] However, most mainstream outdoor tracking cameras on the market currently operate in a standalone mode, meaning each device independently performs all functions, including infrared sensing, image acquisition, and local storage, lacking communication and collaboration capabilities between cameras. In practical ecological monitoring applications, researchers typically need to deploy dozens or even hundreds of cameras over a large area to cover key animal activity paths or habitat hotspots. Because the data is scattered across the local memory cards of each device, users must manually retrieve, download, and process the data from each camera individually. This process is not only extremely time-consuming and labor-intensive, but also yields information limited to single-point records, resulting in limited content that is insufficient to support systematic analysis needs.

[0004] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main purpose of this application is to provide a method for animal collection and identification using distributed cameras, which aims to solve the technical problems of data dispersion and low efficiency of manual data collection and processing caused by the common use of stand-alone working mode in existing outdoor tracking cameras.

[0006] To achieve the above objectives, this application proposes a method for animal acquisition and identification using distributed cameras, the method comprising: Each distributed node's camera detects changes in infrared radiation intensity in its respective target area using a built-in passive infrared sensor. When the passive infrared sensor of any distributed node detects that the change in infrared radiation intensity of the corresponding target area exceeds a preset threshold, the camera of the corresponding distributed node acquires an image of the target area through its built-in image acquisition module, outputs the corresponding image data, and caches the image data in the local storage module of the corresponding distributed node. The cameras at each distributed node transmit the image data to the gateway via their built-in wireless communication modules at preset time intervals. The gateway receives the image data from each distributed node and inputs the image data from each distributed node into a pre-trained animal recognition model. Combined with spatiotemporal correlation analysis, it generates activity trajectories and behavioral statistics of different types of animals.

[0007] In one embodiment, the step of the cameras of each distributed node transmitting image data to the gateway via a built-in wireless communication module at preset time intervals includes: The cameras at each distributed node send the image data to the Wi-Fi HaLow gateway via the built-in Wi-Fi HaLow module at preset time intervals.

[0008] In one embodiment, the image data includes the image body, the camera's unique device identifier, and the image acquisition timestamp; the gateway pre-stores an animal identification database, and the animal identification model is pre-trained based on the animal identification database to identify different types of target animals in the database; the step of inputting the image data from each distributed node into the pre-trained animal identification model and combining it with spatiotemporal correlation analysis to generate activity trajectories and behavioral statistics of different types of animals includes: The image data uploaded by each distributed node is input into the animal recognition model, and the corresponding recognition results are output. If, based on the recognition result, it is determined that the image body includes the target animal in the animal recognition database, then the corresponding image data is classified into the storage directory corresponding to the type of the target animal; Based on the image data in the corresponding storage directory, obtain the unique device identifier of the camera and the image acquisition timestamp carried in the image data, and combine it with the geographical location information corresponding to the camera with different preset device identifiers to determine the geographical location and time of occurrence of each target animal recognition event. Based on the geographical location and time of each identified event of the target animal, the corresponding activity trajectory and behavioral statistics of the target animal are generated.

[0009] In one embodiment, the step of generating the corresponding target animal's activity trajectory and behavioral statistics based on the geographical location and time of occurrence of each target animal identification event includes: Based on the geographical location and time of occurrence of each identified event of the corresponding target animal, the geographical location of the corresponding target animal is associated with the time series, the movement trajectory of the corresponding target animal in multiple target areas is reconstructed, and the activity frequency, active period and path preference of the corresponding target animal are statistically analyzed to obtain the activity trajectory and behavioral statistics of the corresponding target animal.

[0010] In one embodiment, after the step of generating activity trajectories and behavioral statistics of different species of animals, the method further includes: The gateway pushes the activity trajectories and behavioral statistics of different types of animals to the user terminal through the server.

[0011] In addition, to achieve the above objectives, this application also proposes a distributed camera system, which includes multiple cameras and a gateway, with the multiple cameras distributed in the monitoring area according to multiple preset distribution nodes; The camera includes: A passive infrared sensor is used to detect changes in the intensity of infrared radiation in a target area and output a corresponding infrared sensing signal. The image acquisition module is used to acquire images of the target area and output the corresponding image data; The wireless communication module is used to transmit data; The local storage module is used to cache data; The main control module is configured as follows: When the infrared radiation intensity change exceeds a preset threshold based on the infrared sensing signal, the image acquisition module is controlled to acquire an image. The image data is received and cached in the local storage module. The image data is sent to the gateway via the wireless communication module at preset time intervals. The gateway is configured as follows: The system receives image data from each distribution node and inputs the image data from each distribution node into a pre-trained animal recognition model. Combined with spatiotemporal correlation analysis, it generates activity trajectories and behavioral statistics of different types of animals.

[0012] In one embodiment, the wireless communication module is a Wi-Fi HaLow module, and the gateway is a Wi-Fi HaLow gateway.

[0013] In one embodiment, the image data includes the image body, the camera's unique device identifier, and the image acquisition timestamp; the gateway pre-stores an animal identification database, and the animal identification model is pre-trained based on the animal identification database to identify different types of target animals in the database; the gateway is configured to: The image data uploaded by each distributed node is input into the animal recognition model, and the corresponding recognition results are output. If, based on the recognition result, it is determined that the image body includes the target animal in the animal recognition database, then the corresponding image data is classified into the storage directory corresponding to the type of the target animal; Based on the image data in the corresponding storage directory, obtain the unique device identifier of the camera and the image acquisition timestamp carried in the image data, and combine it with the geographical location information corresponding to the camera with different preset device identifiers to determine the geographical location and time of occurrence of each target animal recognition event. Based on the geographical location and time of each identified event of the target animal, the corresponding activity trajectory and behavioral statistics of the target animal are generated.

[0014] In one embodiment, the gateway is configured as follows: Based on the geographical location and time of occurrence of each identified event of the corresponding target animal, the geographical location of the corresponding target animal is associated with the time series, the movement trajectory of the corresponding target animal in multiple target areas is reconstructed, and the activity frequency, active period and path preference of the corresponding target animal are statistically analyzed to obtain the activity trajectory and behavioral statistics of the corresponding target animal.

[0015] In one embodiment, the distributed camera system further includes: The server is used to receive activity trajectories and behavioral statistics of different types of animals sent by the gateway, and push the activity trajectories and behavioral statistics of different types of animals to the user terminal.

[0016] One or more technical solutions proposed in this application have at least the following technical effects: The distributed camera animal acquisition and identification method provided in this application includes the following steps: Step S10, each distributed node's camera detects changes in infrared radiation intensity in its respective target area using a built-in passive infrared sensor. Step S20, when the passive infrared sensor of any distributed node detects that the change in infrared radiation intensity in the corresponding target area exceeds a preset threshold, the camera of the corresponding distributed node acquires an image of the target area through its built-in image acquisition module, outputs the corresponding image data, and caches the image data in the local storage module of the corresponding distributed node. Step S30, each distributed node's camera sends the image data to the gateway through its built-in wireless communication module at preset time intervals. Step S40, the gateway receives the image data from each distributed node and inputs the image data from each distributed node into a pre-trained animal recognition model. Combined with spatiotemporal correlation analysis, it generates activity trajectories and behavioral statistics data for different types of animals. Thus, compared with existing technologies, this embodiment eliminates the need for manual memory card retrieval and can generate activity trajectories and behavioral statistics data for different types of animals, solving the technical problems of data dispersion and low efficiency in manual data retrieval and processing caused by the common use of stand-alone independent working modes in existing outdoor tracking cameras. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating an embodiment of the animal acquisition and identification method using distributed cameras in this application. Figure 2 This is a flowchart illustrating Embodiment 2 of the animal acquisition and identification method using distributed cameras in this application. Figure 3 This is a schematic diagram of a distributed camera system according to an embodiment of this application.

[0020] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0022] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0023] The main solution of this application embodiment is to provide a distributed camera-based animal acquisition and identification method, which includes: Step S10, where cameras at each distributed node detect changes in infrared radiation intensity in their respective target areas using built-in passive infrared sensors. Step S20, when the passive infrared sensor of any distributed node detects that the change in infrared radiation intensity in the corresponding target area exceeds a preset threshold, the camera at the corresponding distributed node acquires an image of the target area using its built-in image acquisition module, outputs the corresponding image data, and caches the image data in the local storage module of the corresponding distributed node. Step S30, where cameras at each distributed node transmit image data to a gateway via a built-in wireless communication module at preset time intervals. Step S40, where the gateway receives the image data from each distributed node and inputs the image data from each distributed node into a pre-trained animal identification model, combining it with spatiotemporal correlation analysis to generate activity trajectories and behavioral statistics data for different types of animals.

[0024] In this embodiment, for ease of description, the following description uses a distributed camera system as the execution subject. The distributed camera system includes multiple cameras and a gateway. The multiple cameras are distributed in the monitoring area according to multiple preset distribution nodes.

[0025] Currently, most mainstream outdoor tracking cameras on the market operate in a standalone mode, meaning each device independently performs all functions, including infrared sensing, image acquisition, and local storage, lacking communication and collaboration capabilities between cameras. In practical ecological monitoring applications, researchers typically need to deploy dozens or even hundreds of cameras over large areas to cover key animal activity paths or habitat hotspots. Because the data is stored separately on the local memory cards of each device, users must manually retrieve, download, and process the data from each camera individually. This process is not only labor-intensive and time-consuming, but also yields information limited to single-point records, resulting in limited content and making it difficult to support systematic analysis needs.

[0026] This application provides a solution whereby cameras at each distributed node acquire and locally cache images of the target area when the infrared radiation intensity change in the target area exceeds a preset threshold. The image data is then automatically and wirelessly uploaded to a gateway at preset intervals. The gateway then uses a pre-trained animal recognition model and performs spatiotemporal correlation analysis based on the timestamp of each image and the node location, thereby automatically generating cross-camera animal activity trajectories and behavioral statistics. Compared to existing technologies, this application eliminates the need for manual memory card retrieval and can generate activity trajectories and behavioral statistics for different types of animals. It solves the technical problems of data dispersion and low efficiency in manual data retrieval and processing caused by the common single-unit independent operation mode of existing outdoor tracking cameras.

[0027] Based on this, embodiments of this application provide a method for animal acquisition and identification using distributed cameras, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the animal acquisition and identification method using distributed cameras according to this application.

[0028] In this embodiment, the animal acquisition and identification method using distributed cameras includes steps S10 to S40: In step S10, the cameras at each distributed node detect changes in infrared radiation intensity in their respective target areas using built-in passive infrared sensors.

[0029] It should be noted that the distributed nodes are independent camera devices deployed in different geographical locations within the monitoring area. Each node can operate autonomously and communicate with the gateway via a network. Passive Infrared Sensors (PIR) are sensors used to receive infrared thermal radiation emitted by heat sources in the environment. When a moving heat source (such as an animal) enters its field of view (i.e., the target area), it causes a momentary change in the intensity of infrared radiation, thereby triggering the output of an infrared sensing signal. In this step, the PIR sensor detects whether any animals have entered the target area where the camera is located. Because PIR sensors have extremely low power consumption, they can remain on standby for extended periods, enabling energy saving and extended battery life for the camera.

[0030] It should be noted that cameras can be deployed in the monitoring area in a grid-like even distribution, with key deployment in hotspot areas (such as water sources, foraging areas, animal trail intersections, and near dens), or in a layered and mixed deployment (sparse grid or boundary deployment in the outer area, and dense deployment in the core area or along paths). There are no restrictions here.

[0031] Step S20: When the passive infrared sensor of any distribution node detects that the change in infrared radiation intensity of the corresponding target area exceeds a preset threshold, the camera of the corresponding distribution node acquires the image of the target area through the built-in image acquisition module, outputs the corresponding image data, and caches the image data in the local storage module of the corresponding distribution node.

[0032] It should be noted that the preset threshold is the minimum infrared change amplitude set to trigger image acquisition. This is used to filter out minor disturbances and ensure that only significant heat sources (such as animals) will trigger image capture. The image acquisition module includes the camera's CMOS / CCD sensor and associated imaging circuitry, responsible for generating visible light or infrared images. The local storage module includes an SD card or Flash memory for caching image data. In this step, once the PIR sensor detects valid heat source activity (infrared radiation intensity change exceeding the preset threshold), the corresponding camera immediately starts taking a picture and temporarily stores the captured image in local memory, awaiting subsequent uploading.

[0033] In step S30, the cameras of each distributed node send image data to the gateway through their built-in wireless communication modules at preset time intervals.

[0034] It should be noted that the wireless communication module can be a LoRa, NB-IoT, Wi-Fi, 4G / 5G module, etc., used to transmit local data to the gateway.

[0035] In one embodiment, step S30 includes step S31: Each distributed node's camera sends image data to the Wi-Fi HaLow gateway via its built-in Wi-Fi HaLow module at preset time intervals.

[0036] It should be noted that Wi-Fi HaLow is a low-power, long-range Wi-Fi technology based on the IEEE 802.11ah standard. It operates in unlicensed frequency bands below 1 GHz and boasts advantages such as long transmission distance, strong wall penetration, low power consumption, and support for a large number of connected devices. In this embodiment, a wireless communication chip or module with an IEEE 802.11ah protocol stack is integrated into the camera and embedded in the distributed node cameras to implement HaLow communication functionality. The Wi-Fi HaLow gateway is an intelligent gateway device integrating Wi-Fi HaLow wireless access functionality and edge AI computing capabilities. It can act as the central node of the wireless network, receiving image data from various distributed camera nodes. It can also perform edge computing, offloading data processing tasks from the cloud to the network edge closer to the data source (i.e., the Wi-Fi HaLow gateway), reducing transmission latency, bandwidth consumption, and dependence on cloud services. The preset time interval is the data upload cycle pre-configured by the camera (e.g., every 6 hours or 2 AM daily).

[0037] In this embodiment, after each distributed node's smart camera completes image acquisition and local caching, it wakes up its built-in Wi-Fi HaLow module at preset time intervals. The cached image data is then transmitted to the Wi-Fi HaLow gateway via a Sub-1GHz Wi-Fi HaLow wireless link. Compared to traditional Wi-Fi (2.4 / 5GHz), Wi-Fi HaLow's Sub-1GHz signal can cover hundreds of meters to over 1 kilometer, making it suitable for large-scale deployment in outdoor environments. This reduces the number of relay nodes and lowers deployment costs. Furthermore, Wi-Fi HaLow constructs a self-organizing private local area network, independent of carrier networks, allowing for flexible deployment.

[0038] In step S40, the gateway receives image data from each distribution node and inputs the image data from each distribution node into a pre-trained animal recognition model. Combined with spatiotemporal correlation analysis, it generates activity trajectories and behavioral statistics of different types of animals.

[0039] It should be noted that the pre-trained animal recognition model is an AI model trained on a large number of wildlife images based on deep learning (such as YOLO, ResNet, etc.), used to automatically identify animal species in the images. Through spatiotemporal correlation analysis, the same animal detected at different times and locations (i.e., different nodes) can be associated to infer its movement path (trajectory) and behavioral patterns (such as nocturnal activity, flocking, etc.). In this step, after the gateway collects the images uploaded by all nodes, it uses the AI ​​model to automatically identify the animal species and combines the image timestamps and geographical location information to reconstruct the animal's movement path and behavioral patterns.

[0040] Thus, compared with the existing technology, this embodiment does not require manual recycling of the memory card and can generate activity trajectories and behavioral statistics of different kinds of animals, solving the technical problems of data dispersion and low efficiency of manual data recycling and processing caused by the common use of stand-alone working mode in existing outdoor tracking cameras.

[0041] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in Embodiment 1 above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 The image data includes the image itself, the camera's unique device identifier, and the image acquisition timestamp; the gateway pre-stores an animal identification database, and the animal identification model is pre-trained based on this database to identify different types of target animals. Step 40 includes steps S41~S44: Step S41: Input the image ontology from the image data uploaded by each distributed node into the animal recognition model and output the corresponding recognition results.

[0042] Step S42: If, based on the recognition result, it is determined that the image ontology includes the target animal in the animal recognition database, then the corresponding image data is classified into the storage directory corresponding to the type of the target animal.

[0043] Step S43: Based on the image data in the corresponding storage directory, obtain the unique device identifier of the camera and the image acquisition timestamp carried in the image data, and combine the preset geographical location information corresponding to the camera with different device identifiers to determine the geographical location and time of occurrence of each target animal identification event.

[0044] Step S44: Based on the geographical location and time of occurrence of each identification event of the target animal, generate the corresponding activity trajectory and behavioral statistics of the target animal.

[0045] In one embodiment, step S44 specifically includes step S441: Based on the geographical location and time of occurrence of each identified event of the corresponding target animal, the geographical location of the corresponding target animal is associated with the time series, the movement trajectory of the corresponding target animal in multiple target areas is reconstructed, and the activity frequency, active period and path preference of the corresponding target animal are statistically analyzed to obtain the activity trajectory and behavioral statistics of the corresponding target animal.

[0046] In this embodiment, the animal identification database can include animals such as wild boar, bear, deer, wolf, fox, bird, cat, and dog. Taking foxes and deer as examples, the two have significant and stable visual differences in morphological structure and appearance characteristics: foxes are usually medium-sized, with a pointed muzzle, large erect ears, a long, bushy tail often with a white tip, and their body color is mostly reddish-brown; deer are larger, with long limbs, a slender neck, and some species (such as sika deer) have obvious white spots on their fur in summer, and males often have forked antlers. The pre-trained animal identification model learns the discriminative features of target species such as foxes and deer in terms of outline proportions, fur color distribution, and limb posture through a large number of labeled samples. Based on the image ontology in the image data, it can achieve high-accuracy automatic identification in complex wild scenes and output high-confidence identification results.

[0047] Then, based on the recognition results, the gateway categorizes the image data into the storage directory corresponding to the species of the target animal. It should be noted that if the image data contains multiple target animals, it is categorized into the respective storage directories for each target animal; if the image data does not contain any animals in the animal identification database, it is either discarded or categorized into the "unrecognized" image directory. This allows users to focus on analyzing a specific subset of data related to a particular target animal.

[0048] It should be noted that the unique device identifier for a camera can be a MAC address, serial number, or custom ID, used to uniquely identify a specific distributed node camera. The image acquisition timestamp is the precise time the image was captured, and the geographic location information is the GPS coordinates preset when each camera node was deployed. In this embodiment, for the image data under each animal category, the unique ID of the source camera is extracted, the known geographic location corresponding to that ID is queried, and combined with the image acquisition timestamp, a complete "identification event" record (species, latitude and longitude, time) is obtained. Finally, based on the geographic location and occurrence time of each identification event for the corresponding target animal, each identification event is sorted according to time sequence, and adjacent events are associated based on spatial continuity to reconstruct the movement trajectory of the target animal within the coverage area of ​​multiple distributed nodes. At the same time, its frequency of occurrence, active periods, and path usage preferences in different areas are statistically analyzed to generate corresponding target animal activity trajectory and behavioral statistics. In this way, based on image data under different storage directories, activity trajectory and behavioral statistics for different types of target animals can be obtained.

[0049] Thus, this embodiment can generate the dynamic behavior of animals based on the point records of cameras at each distributed node, and obtain the activity trajectories and behavior statistics of different types of animals in the corresponding animal identification database, which helps to reveal the real activity patterns of animals in the animal identification database.

[0050] Based on the first embodiment of this application, in the third embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description and will not be repeated hereafter. In addition, the animal acquisition and identification method using distributed cameras further includes step S50: The gateway pushes the activity trajectories and behavioral statistics of different types of animals to the user terminal through the server.

[0051] In this embodiment, the server can automatically push the activity trajectories and behavioral statistics of different animal species to the user terminal, allowing the user to promptly grasp the activity trajectories and behavioral patterns of different animal species. This enables users to detect abnormal activities of specific animal species in a timely manner, thus improving the user experience.

[0052] This application also provides a distributed camera system, as shown in the reference. Figure 3 The distributed camera system includes multiple cameras and gateways, with the cameras distributed across multiple preset distribution nodes in the monitored area. The camera includes: A passive infrared sensor is used to detect changes in the intensity of infrared radiation in a target area and output a corresponding infrared sensing signal. The image acquisition module is used to acquire images of the target area and output the corresponding image data; The wireless communication module is used to transmit data; The local storage module is used to cache data; The main control module is configured as follows: When the infrared radiation intensity change exceeds a preset threshold based on the infrared sensing signal, the image acquisition module is controlled to acquire an image. Receive image data and cache the image data in the local storage module; The image data is sent to the gateway via the wireless communication module at preset time intervals. The gateway is configured as follows: The system receives image data from each distribution node and inputs it into a pre-trained animal recognition model. Combined with spatiotemporal correlation analysis, it generates activity trajectories and behavioral statistics of different types of animals.

[0053] It should be noted that the multiple preset distribution nodes of each camera can be deployed in the monitoring area in the following ways: evenly distributed in a grid, focused on hot spots (such as water sources, foraging areas, animal trail intersections, and near dens), or layered and mixed deployment (sparse grid or boundary deployment in the outer area, and dense deployment in the core area, such as hot spots or paths). There are no restrictions here.

[0054] In this embodiment, the passive infrared sensor (PIR) is used to receive infrared thermal radiation emitted by heat sources in the environment. When a moving heat source (such as an animal) enters its field of view (i.e., the target area), it causes a momentary change in the intensity of infrared radiation, thereby triggering the output of an infrared sensing signal. The image acquisition module includes the camera's CMOS / CCD sensor and its associated imaging circuitry, responsible for generating visible light or infrared images. The local storage module includes an SD card or Flash memory for caching image data. The main control module can be a microcontroller or an embedded SOC, used to monitor the output of the PIR sensor. When the change in the infrared signal exceeds a preset threshold (indicating the entry of a target), image acquisition is immediately initiated, and the acquired image is stored in the local storage module. The image is then uploaded to the gateway at preset time intervals (e.g., every 6 hours or 2 AM daily) to save energy and bandwidth.

[0055] The distributed camera system provided in this embodiment enables cameras at each distributed node to capture and locally cache images of the target area when the infrared radiation intensity change in the target area exceeds a preset threshold. Subsequently, the image data is automatically and wirelessly uploaded to a gateway at preset intervals. The gateway then uniformly calls a pre-trained animal recognition model and performs spatiotemporal correlation analysis based on the timestamp of each image and the node location, thereby automatically generating cross-camera animal activity trajectories and behavioral statistics. Compared with existing technologies, this application eliminates the need for manual memory card retrieval and can generate activity trajectories and behavioral statistics for different types of animals, solving the technical problems of data dispersion and low efficiency in manual data retrieval and processing caused by the common single-machine independent working mode of existing outdoor tracking cameras.

[0056] In one embodiment, the wireless communication module is a Wi-Fi HaLow module, and the gateway is a Wi-Fi HaLow gateway.

[0057] In this embodiment, the Wi-Fi HaLow module and the Wi-Fi HaLow gateway communicate via a Sub-1GHz signal. This Sub-1GHz signal can cover hundreds of meters to over 1 kilometer, making it suitable for large-scale deployment in outdoor environments. This reduces the number of relay nodes and lowers deployment costs. Furthermore, Wi-Fi HaLow constructs a self-organizing private local area network, independent of carrier networks, allowing for flexible deployment.

[0058] In one embodiment of this application, the image data includes the image body, a unique device identifier for the camera, and an image acquisition timestamp; the gateway pre-stores an animal identification database, and the animal identification model is pre-trained based on the animal identification database to identify different types of target animals in the database; the gateway is configured as follows: The image ontology from the image data uploaded by each distributed node is input into the animal recognition model, and the corresponding recognition results are output. If, based on the recognition results, it is determined that the image ontology includes the target animal in the animal recognition database, then the corresponding image data will be classified into the storage directory corresponding to the species of the target animal. Based on the image data in the corresponding storage directory, obtain the unique device identifier of the camera and the image acquisition timestamp carried in the image data. Combined with the geographical location information corresponding to the camera with different preset device identifiers, determine the geographical location and time of occurrence of each target animal identification event. Based on the geographical location and time of each identified event of the target animal, the corresponding activity trajectory and behavioral statistics of the target animal are generated.

[0059] In one implementation, the gateway is specifically configured as follows: Based on the geographical location and time of occurrence of each identified event of the corresponding target animal, the geographical location of the corresponding target animal is associated with the time series, the movement trajectory of the corresponding target animal in multiple target areas is reconstructed, and the activity frequency, active period and path preference of the corresponding target animal are statistically analyzed to obtain the activity trajectory and behavioral statistics of the corresponding target animal.

[0060] In this embodiment, the animal identification database can include animals such as wild boar, bear, deer, wolf, fox, bird, cat, and dog. Taking foxes and deer as examples, the two have significant and stable visual differences in morphological structure and appearance characteristics: foxes are usually medium-sized, with a pointed muzzle, large erect ears, a long, bushy tail often with a white tip, and their body color is mostly reddish-brown; deer are larger, with long limbs, a slender neck, and some species (such as sika deer) have obvious white spots on their fur in summer, and males often have forked antlers. The pre-trained animal identification model learns the discriminative features of target species such as foxes and deer in terms of outline proportions, fur color distribution, and limb posture through a large number of labeled samples. Based on the image ontology in the image data, it can achieve high-accuracy automatic identification in complex wild scenes and output high-confidence identification results.

[0061] Then, based on the recognition results, the gateway categorizes the image data into the storage directory corresponding to the species of the target animal. It should be noted that if the image data contains multiple target animals, it is categorized into the respective storage directories for each target animal; if the image data does not contain any animals in the animal identification database, it is either discarded or categorized into the "unrecognized" image directory. This allows users to focus on analyzing a specific subset of data related to a particular target animal.

[0062] It should be noted that the unique device identifier for a camera can be a MAC address, serial number, or custom ID, used to uniquely identify a specific distributed node camera. The image acquisition timestamp is the precise time the image was captured, and the geographic location information is the GPS coordinates preset when each camera node was deployed. In this embodiment, for the image data under each animal category, the unique ID of the source camera is extracted, the known geographic location corresponding to that ID is queried, and combined with the image acquisition timestamp, a complete "identification event" record (species, latitude and longitude, time) is obtained. Finally, based on the geographic location and occurrence time of each identification event for the corresponding target animal, each identification event is sorted according to the time series, and adjacent events are associated based on spatial continuity to reconstruct the movement trajectory of the target animal within the coverage area of ​​multiple distributed nodes. At the same time, the frequency of its appearance, active periods, and path usage preferences in different areas are statistically analyzed to generate the corresponding target animal's activity trajectory and behavioral statistics.

[0063] Thus, this embodiment can generate the dynamic behavior of animals based on the point records of cameras at each distributed node, obtain the activity trajectory and behavior statistics of animals in the animal identification database, and help to reveal the real activity patterns of animals in the animal identification database.

[0064] In one embodiment of this application, the distributed camera system further includes: The server is used to receive activity trajectories and behavioral statistics of different types of animals sent by the gateway, and push the activity trajectories and behavioral statistics of different types of animals to the user terminal.

[0065] In this embodiment, the server can automatically push the activity trajectories and behavioral statistics of different animal species to the user terminal, allowing the user to promptly grasp the activity trajectories and behavioral patterns of different animal species. This enables users to detect abnormal activities of specific animal species in a timely manner, thus improving the user experience.

[0066] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for animal acquisition and identification using distributed cameras, characterized in that, The method includes: Each distributed node's camera detects changes in infrared radiation intensity in its respective target area using a built-in passive infrared sensor. When the passive infrared sensor of any distributed node detects that the change in infrared radiation intensity of the corresponding target area exceeds a preset threshold, the camera of the corresponding distributed node acquires an image of the target area through its built-in image acquisition module, outputs the corresponding image data, and caches the image data in the local storage module of the corresponding distributed node. The cameras at each distributed node transmit the image data to the gateway via their built-in wireless communication modules at preset time intervals. The gateway receives the image data from each distributed node and inputs the image data from each distributed node into a pre-trained animal recognition model. Combined with spatiotemporal correlation analysis, it generates activity trajectories and behavioral statistics of different types of animals.

2. The method as described in claim 1, characterized in that, The step of each distributed node's camera sending image data to the gateway via its built-in wireless communication module at preset time intervals includes: The cameras at each distributed node send the image data to the Wi-Fi HaLow gateway via the built-in Wi-Fi HaLow module at preset time intervals.

3. The method as described in claim 1, characterized in that, The image data includes the image itself, the camera's unique device identifier, and the image acquisition timestamp; the gateway pre-stores an animal identification database, and the animal identification model is pre-trained based on the animal identification database to identify different types of target animals in the database; the step of inputting the image data from each distributed node into the pre-trained animal identification model, and combining it with spatiotemporal correlation analysis to generate activity trajectories and behavioral statistics of different types of animals includes: The image data uploaded by each distributed node is input into the animal recognition model, and the corresponding recognition results are output. If, based on the recognition result, it is determined that the image body includes the target animal in the animal recognition database, then the corresponding image data is classified into the storage directory corresponding to the type of the target animal; Based on the image data in the corresponding storage directory, obtain the unique device identifier of the camera and the image acquisition timestamp carried in the image data, and combine the preset geographical location information corresponding to cameras with different device identifiers to determine the geographical location and time of occurrence of each target animal recognition event. Based on the geographical location and time of each identified event of the target animal, the corresponding activity trajectory and behavioral statistics of the target animal are generated.

4. The method as described in claim 3, characterized in that, The step of generating the corresponding target animal's activity trajectory and behavioral statistics based on the geographical location and time of occurrence of each target animal identification event includes: Based on the geographical location and time of occurrence of each identified event of the corresponding target animal, the geographical location of the corresponding target animal is associated with the time series, the movement trajectory of the corresponding target animal in multiple target areas is reconstructed, and the activity frequency, active period and path preference of the corresponding target animal are statistically analyzed to obtain the activity trajectory and behavioral statistics of the corresponding target animal.

5. The method as described in claim 1, characterized in that, Following the step of generating activity trajectories and behavioral statistics for different species of animals, the method further includes: The gateway pushes the activity trajectories and behavioral statistics of different types of animals to the user terminal through the server.

6. A distributed camera system, characterized in that, It includes multiple cameras and gateways, with the cameras distributed across multiple preset distribution nodes in the monitored area; The camera includes: A passive infrared sensor is used to detect changes in the intensity of infrared radiation in a target area and output a corresponding infrared sensing signal. The image acquisition module is used to acquire images of the target area and output the corresponding image data; The wireless communication module is used to transmit data; The local storage module is used to cache data; The main control module is configured as follows: When the infrared radiation intensity change exceeds a preset threshold based on the infrared sensing signal, the image acquisition module is controlled to acquire an image. The image data is received and cached in the local storage module. The image data is sent to the gateway via the wireless communication module at preset time intervals. The gateway is configured as follows: The system receives image data from each distributed node and inputs the image data from each distributed node into a pre-trained animal recognition model. Combined with spatiotemporal correlation analysis, it generates activity trajectories and behavioral statistics of different types of animals.

7. The distributed camera system as described in claim 6, characterized in that, The wireless communication module is a Wi-Fi HaLow module, and the gateway is a Wi-Fi HaLow gateway.

8. The distributed camera system as described in claim 6, characterized in that, The image data includes the image itself, the camera's unique device identifier, and the image acquisition timestamp; the gateway pre-stores an animal identification database, and the animal identification model is pre-trained based on this database to identify different types of target animals; the gateway is configured as follows: The image data uploaded by each distributed node is input into the animal recognition model, and the corresponding recognition results are output. If, based on the recognition result, it is determined that the image body includes the target animal in the animal recognition database, then the corresponding image data is classified into the storage directory corresponding to the type of the target animal; Based on the image data in the corresponding storage directory, obtain the unique device identifier of the camera and the image acquisition timestamp carried in the image data, and combine the preset geographical location information corresponding to cameras with different device identifiers to determine the geographical location and time of occurrence of each target animal recognition event. Based on the geographical location and time of each identified event of the target animal, the corresponding activity trajectory and behavioral statistics of the target animal are generated.

9. The distributed camera system as described in claim 8, characterized in that, The gateway is configured as follows: Based on the geographical location and time of occurrence of each identified event of the corresponding target animal, the geographical location of the corresponding target animal is associated with the time series, the movement trajectory of the corresponding target animal in multiple target areas is reconstructed, and the activity frequency, active period and path preference of the corresponding target animal are statistically analyzed to obtain the activity trajectory and behavioral statistics of the corresponding target animal.

10. The distributed camera system as described in claim 6, characterized in that, Also includes: The server is used to receive activity trajectories and behavioral statistics of different types of animals sent by the gateway, and push the activity trajectories and behavioral statistics of different types of animals to the user terminal.

Images