Computer-implemented systems and methods for tracking objects in events

Abstract

Computer-implemented systems and methods are provided for tracking objects. In one implementation, a method for tracking objects engaged in an event within a venue is provided. The method may be implemented with at least one processor configured with executable instructions to perform steps including: providing a statistical model based on historical event data for at least one event type; providing a 3-dimensional (3D) model of the venue using map data; determining a trajectory within the venue based on the 3D and statistical models; and determining a sequence of image data capture for a set of cameras installed at the venue based on the determined trajectory. The method may also include receiving, via an external controller, one or more commands to modify automatic tracking, and generating a broadcast feed comprising the captured image data.

Description

COMPUTER-IMPLEMENTED SYSTEMS AND METHODS FOR TRACKINGOBJECTS IN EVENTSTECHNICAL FIELD

[0001] The present disclosure relates generally to computer-implemented systems and methods for tracking objects in events, such as objects engaged in events within a racing track, stadium, or other venue. More specifically, and without limitation, this disclosure relates to automatically tracking objects in an event by utilizing one or more of historical event data, 3D model(s) for the venue, and / or statistical model(s). Consistent with the disclosed embodiments, non-transitory computer-readable storage media may store program instructions, which may be executable by at least one processing device and perform any of the steps and / or methods described herein.BACKGROUND

[0002] In the dynamic landscape of event management and activities, including within the realm of racing events, sports competitions, and large-scale venues, the precise tracking of objects, including participants, individuals, and / or other objects in an event, remains an important challenge. The objects engaged in the event may be moving within or across the venue and tracking them may be required for filming, recording, capturing and relaying data, or broadcast. Traditional methods, including those reliant on manual control and / or rudimentary algorithms, often fall short of providing the level of accuracy and automation necessary for efficient and effective event monitoring, tracking, and analysis.

[0003] Extant solutions for tracking objects include systems and methods that are predominately reliant on human operation or rooted in computer-vision technologies.Human operators are expensive, unreliable, and increasingly hard to find. Human operated cameras can also be overly complex for a required project. On the other hand, computer-vision systems tend to be computationally expensive and / or prone to error. In addition, when tracking is attempted predominately through a computer-vision system, latency is added to the processing chain as computations are undertaken by the system and when the output is forwarded to other components. Robotic Pan-Tilt-Zoom (PTZ) camera systems have been developed and may provide near-field tracking capabilities to frame a single object. However, these systems are limited in scope and may require manual adjustments by skilled operators for optimal performance. Fixed camera systems with wide-angle views may be suitable for capturing objects and event-related activities across larger arenas, but still require human operators, which can introduce potential errors and significant costs.

[0004] The limitations of extant approaches are further compounded by the inherent complexities of these systems. For example, for computer-vision systems, factors such as incomplete or inaccurate data, fluctuating light levels, contrast discrepancies, and adverse weather conditions can significantly impede object recognition accuracy leading to false positives and negatives in image recognition. Moreover, challenges such as object obfuscation within the field of view and the need for reliable object re-identification underscore the need for an improved tracking solution.

[0005] In this context, there arises a demand for innovative object tracking systems and methods that offer enhanced accuracy and automation while mitigating one or more of the shortcomings inherent in extant solutions and methodologies. For example, there is a need for improved tracking solutions that are more reliable, efficient, and less dependent on human intervention or computer-vision systems. Solutions are also needed that can provide real-time object tracking for live events and complex scenes or conditions. The present disclosure describes solutions to alleviate or overcome one ormore of the above-stated problems, among others with conventional systems and methods.SUMMARY

[0006] Embodiments consistent with the present disclosure provide computer- implemented systems and methods for tracking objects within a venue, such as participants, individuals, and / or other objects engaged in events. Embodiments of the present disclosure may utilize one or more statistical and 3-dimensional (3D) models. The disclosed embodiments may be implemented using a combination of computer hardware and software, as well as specialized hardware, software, and / or firmware.

[0007] In an embodiment, a computer-implemented method is provided for tracking objects engaged in an event within a venue. The method may comprise: providing a statistical model based on historical event data for at least one event type; providing a 3- dimensional (3D) model of the venue using map data; determining a trajectory within the venue based on the 3D and statistical models; and determining a sequence of image data capture for a set of cameras installed at the venue based on the determined trajectory, wherein the sequence enables one or more objects engaged in an event to be automatically tracked along the determined trajectory via the set of cameras installed at the venue, and wherein image data of the one or more objects are captured by the set of cameras in the determined sequence. The method may also include receiving, via an external controller, one or more commands to modify the automatic tracking. Further, the method may include generating a broadcast feed comprising the captured image data.

[0008] In another embodiment, a non-transitory computer-readable medium is provided that contains instructions that when executed by at least one processor causes the at least one processor to perform operations for tracking objects engaged in an event within a venue. The operations may comprise: providing a statistical model based onhistorical event data for at least one event type; providing a 3-dimensional (3D) model of the venue using map data; determining a trajectory within the venue based on the 3D and statistical models; and determining a sequence of image data capture for a set of cameras installed at the venue based on the determined trajectory, wherein the sequence enables one or more objects engaged in an event to be automatically tracked along the determined trajectory via the set of cameras installed at the venue, and wherein image data of the one or more objects are captured by the set of cameras in the determined sequence . The operations may also include receiving, via an external controller, one or more commands to modify the automatic tracking. Further, the operations may include generating a broadcast feed comprising the captured image data.

[0009] In a further embodiment, a computer-implemented system is provided for tracking objects engaged in an event within a venue. The system may include at least one processor configured with executable instructions to: provide a statistical model based on historical event data for at least one event type; provide a 3-dimensional (3D) model of the venue using map data; determine a trajectory within the venue based on the 3D and statistical models; and determine a sequence of image data capture for a set of cameras installed at the venue based on the determined trajectory, wherein the sequence enables one or more objects engaged in an event to be automatically tracked along the determined trajectory via a set of cameras installed at the venue, and wherein image data of the one or more objects are captured by the set of cameras in a sequence determined based on the trajectory. The processor may also be configured to receive, via an external controller, one or more commands to modify the automatic tracking. Further, the processor may be configured to generate a broadcast feed comprising the captured image data.

[0010] The foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the claims. Other embodiments and features may be implemented, consistent with the present disclosure.BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate various disclosed embodiments. In the drawings:

[0012] FIG. 1A is a diagrammatic representation of an exemplary system for tracking objects, consistent with the disclosed embodiments;

[0013] FIG. 1 B is a diagrammatic representation of another exemplary system for tracking objects, consistent with the disclosed embodiments;

[0014] FIG. 2A is a flowchart showing an exemplary initialization process for tracking objects, consistent with the disclosed embodiments;

[0015] FIG. 2B is a flowchart showing an exemplary process for tracking objects engaged in an event within a venue, consistent with the disclosed embodiments;

[0016] FIG. 3 is an illustration of an exemplary venue, consistent with the disclosed embodiments;

[0017] FIG. 4A is an illustration of an exemplary 3D model of a track and field venue, consistent with the disclosed embodiments;

[0018] FIG. 4B is an illustration of an exemplary 3D model of an auto racing circuit, consistent with the disclosed embodiments;

[0019] FIG. 5 is an exemplary representation of an outcome of a statistical model based on historical event data combined with a 3D model, consistent with the disclosed embodiments;

[0020] FIG. 6 is an illustration of an exemplary combined 3D and statistical model featuring a set of cameras, consistent with the disclosed embodiments;

[0021] FIG. 7 is an illustration of an exemplary coverage of the set of cameras illustrated in FIG. 6, consistent with the disclosed embodiments;

[0022] FIG. 8A is a diagrammatic representation of a first exemplary system configuration, consistent with the disclosed embodiments; and

[0023] FIG. 8B is a diagrammatic representation of a second exemplary system configuration, consistent with the disclosed embodiments.DETAILED DESCRIPTION

[0024] The following detailed description refers to the accompanying drawings. Wherever possible, the same reference numbers are used in the drawings and the following description to refer to the same or similar parts. While several illustrative embodiments are described herein, modifications, adaptations and other implementations are possible. For example, substitutions, additions or modifications may be made to the components illustrated in the drawings, and the illustrative methods described herein may be modified by substituting, reordering, removing, or adding steps to the disclosed methods. Accordingly, the following detailed description is not limited to the disclosed embodiments and examples. Instead, the proper scope is defined by the appended claims. Also, it is to be understood that the present disclosure may be practiced without one or more of these details.

[0025] Embodiments described herein may refer to a non-transitory computer- readable medium containing instructions that when executed by at least one processor, cause the at least one processor to perform a method. Non-transitory computer-readable mediums may be any medium capable of storing data in any memory in a way that may be read by any computing device with a processor to carry out methods or any other instructions stored in the memory. The non-transitory computer-readable medium may be implemented as hardware, firmware, software, or any combination thereof. Moreover, thesoftware may preferably be implemented as an application program tangibly embodied on a program storage unit or computer-readable medium consisting of parts, or of certain devices and / or a combination of devices. The application program may be uploaded to, and executed by, a machine comprising any suitable architecture. Preferably, the machine may be implemented on a computer platform having hardware such as one or more central processing units (“CPUs”), a memory, and input / output interfaces. The computer platform may also include an operating system and microinstruction code. The various processes and functions described in this disclosure may be either part of the microinstruction code or part of the application program, or any combination thereof, which may be executed by a CPU, whether or not such a computer or processor is explicitly shown. In addition, various other peripheral units may be connected to the computer platform such as an additional data storage unit and a printing unit. Furthermore, a non-transitory computer-readable medium may be any computer- readable medium except for a transitory propagating signal.

[0026] The memory may include a Random Access Memory (RAM), a Read-Only Memory (ROM), a hard disk, an optical disk, a magnetic medium, a flash memory, other permanent, fixed, volatile or non-volatile memory, or any other mechanism capable of storing instructions. The memory may include one or more separate storage devices collocated or disbursed, capable of storing data structures, instructions, or any other data. The memory may further include a memory portion containing instructions for the processor to execute. The memory may also be used as a working scratch pad for the processors or as a temporary storage.

[0027] Some embodiments may involve at least one processor. A processor may be any physical device or group of devices having electric circuitry that performs a logic operation on input or inputs. For example, the at least one processor may include one or more integrated circuits (IC), including application-specific integrated circuit (ASIC),microchips, microcontrollers, microprocessors, all or part of a central processing unit (CPU), graphics processing unit (GPU), digital signal processor (DSP), field- programmable gate array (FPGA), server, virtual server, or other circuits suitable for executing instructions or performing logic operations. The instructions executed by at least one processor may, for example, be pre-loaded into a memory integrated with or embedded into the controller or may be stored in a separate memory.

[0028] In some embodiments, the at least one processor may include more than one processor. Each processor may have a similar construction, or the processors may be of differing constructions that are electrically connected or disconnected from each other. For example, the processors may be separate circuits or integrated in a single circuit. When more than one processor is used, the processors may be configured to operate independently or collaboratively. The processors may be coupled electrically, magnetically, optically, acoustically, mechanically or by other means that permit them to interact.

[0029] Embodiments of the present disclosure include computer-implemented systems and methods for tracking objects engaged in an event within a venue, such as participants, individuals, and / or other objects in the event. The disclosed embodiments may be used for automating the capture of various types of events (e.g., track and field races, automotive and motoring races, horse and greyhound races, etc.) that take place along a determined trajectory, between defined start and end positions. The disclosed embodiments may be implemented using 3D model(s) for the venue and / or statistical model(s) based on historical event data. One or more cameras may be installed in the venue and mapped into the 3D model or combined 3D and statistical model. The cameras and / or other image capturing devices may capture and provide image data. As used herein, “image data” includes still images and videos. Still images are single, static images captures at a particular moment in time. Videos contain multiple still images orframes that can be played back in succession to provide a full motion capture of an event, for example.

[0030] As disclosed herein, 3D model(s) may be constructed using map data (e.g., GPS data, engineering drawings, site surveys, etc.). In some embodiments, a 3D model may include or be combined with a statistical model that is built using historical event data for a given event type. The historical event data for an event type may include data such as timing / acceleration characteristics and other data from prior events. In some embodiments, the statistical model may incorporate at least the event duration and average speeds / accelerations of objects along the path or trajectory for the event. Specific event types can also be modeled to include scenes from a pre-event, start, features of interest, finish, post-event, etc. The 3D and statistical models can be used to determine a trajectory and automate the tracking of objects engaged in an event within the venue. For a given event, the tracking of the objects can be further refined using an external human operator and / or programmed control system. In some embodiments, the distances or points along the trajectory may be used to define steps in the video camera capture sequence and / or recording workflow.

[0031] Various embodiments and features of the invention are presented herein. By way of example, embodiments of the invention include computer-implemented systems and methods for tracking objects engaged in an event within a venue, wherein a 3D model is provided of the venue and created using map data from one or more sources such as GPS data, engineering drawings, dedicated site surveys, etc. The 3D model may include or be combined a statistical model which is built using historical event data for a given event type. The statistical model may include characteristics such as event duration, average speeds, and average accelerations for the objects. In some embodiments, one or more cameras may be installed at the venue and mapped into the 3D model or combined 3D and statistical model. The cameras may provide video or image feedsduring the event. The mapping may include mapping the positions of the cameras into a 3D coordinate system and mapping key event locations such as pre-event, start, features of interest, finish, and post-finish. In some embodiments, the 3D and statistical models are used to automate the tracking and recording of objects engaged in an event by predicting their movement along a determined path or trajectory for the duration of the event and controlling the selection and / or sequence of camera feeds. An interface may be provided to allow an operator to interact with the system (e.g., to trigger the start of tracking an event or to make tracking adjustments during the event). Further, the distances or points along the determined trajectory may be used to recall parameters and / or metadata relevant to tracking and recording. For example, the distance along the trajectory may be used to define steps in the video capture and / or recording workflow. During the capture of an event, an operator or other user may supplement the automated tracking of the objects. This may be done via an external controller to, for example, optimize the framing of the objects engaged in the event and / or the capture of images or video.

[0032] Consistent with the embodiments of this disclosure, “tracking objects” refers to an operation or process capable of ensuring that the positions of one or more objects engaged in an event are captured by at least one camera. Such tracking may include the systematic and continuous tracking of objects or subjects as they move along a determined path or trajectory. Further, tracking objects may involve the coordinated operation of one or more cameras strategically placed in a venue to maintain visual surveillance and capture the movements of the objects or subjects of interest engaged in the event. As disclosed herein, 3D and statistical models may be used to automate the tracking and recording of objects engaged in an event by predicting their movement along a determined trajectory for the duration of the event and controlling the selection and / or sequence of feeds from one or more cameras and / or other image capturing devices.

[0033] FIG. 1A is a diagrammatic representation of an exemplary system 100 for tracking objects, consistent with the disclosed embodiments. System 100 may include various components depending on the requirements of a particular implementation. As shown in the example of FIG. 1A, system 100 may include at least one processor or a processing unit 110. In some embodiments, processing unit 110 may include an applications processor, an image processor, and / or any other suitable processing device. Processing unit 110 may be communicatively connected to one or more devices and other components, such as a switcher 150 and an external controller 170. Communication between processing unit 110 and the switcher 150, as well as between processing unit 110 and the external controller 170, or between processing unit 110 and any other suitable devices or components, may be performed via one or more communication links, networks, or means. In some embodiments, the one or more communication links, networks, or means may include wired connections, epitomized by standards such as USB, Ethernet, and HDMI, to offer reliable and high-bandwidth links suitable for tasks ranging from data transfer to multimedia streaming. These connections leverage physical cables to establish direct and robust communication pathways between the processing unit and connected devices, with minimal latency and interference. In some embodiments, the one or more communication links, networks, or means may include wireless communication technologies, such as Wi-Fi, Bluetooth, Zigbee, cellular networks, and NFC, to liberate devices from the constraints of physical cables, fostering greater mobility and flexibility in various configurations. Further, in some embodiments, the one or more communication links, networks, or means may include a combination of wired connections and wireless communication technologies possibly involving one or more communication relay devices to facilitate seamless communication across a network. Communication via the one or more communication links, networks, or means may be plain and unencrypted or encrypted.

[0034] Processing unit 110 of FIG. 1A may be configured to perform certain operations or functions, including for tracking objects engaged in an event. Configuring processing unit 110 or other processing devices, including a processor or other controller or microprocessor, to perform certain operations or functions may include providing computer-executable instructions and making those instructions available to the processing unit or device for execution by the processing unit or device. In some embodiments, configuring a processing unit or device may include programming the processing unit or device directly with firmware or architectural instructions. In other embodiments, configuring a processing unit or device may include storing executable instructions on a memory that is accessible to the processing unit or device during operation. For example, the processing unit or device may access the memory to obtain and execute the stored instructions during operation.

[0035] Processing unit 110 may comprise various types of devices and / or functions, implemented through any combination of hardware, software, and / or firmware. For example, processing unit 110 may include a controller, a central processing unit (CPU), support circuits, digital signal processors, integrated circuits, memory, or any other types of devices. The CPU may comprise any number of processors or microcontrollers. The support circuits may be any number of circuits generally well-known in the art, including cache, power supply, clock and input-output circuits. The memory associated with processing unit 110 may store software or instructions that, when executed by the processor, control the operation of the processor. The memory may include databases, image processing software, and three-dimensional graphics processing software, as well as a trained system, such as a neural network, or a deep neural network, for example. The memory may comprise any number of random-access memories, read only memories, flash memories, disk drives, optical storage, tape storage, removable storage and other types of storage. In one instance, the memory may be separate from theprocessing unit 110. In another instance, the memory may be integrated into the processing unit 110. In some embodiments, the memory may receive and store data, including historical event data 125 and map data 135. The memory may also store one or more statistical models 120 generated from the historical event data 125 and one or more 3-dimensional (3D) models 130 generated from the map data 135.

[0036] As further disclosed herein, 3D and statistical models can be used to automate the tracking of objects engaged in an event within a venue. A 3D model 130 of a venue may be constructed using map data 135, such as GPS data, engineering drawings, site surveys, etc. A path or trajectory for an event within a venue may be determined from one or more of the 3D model(s) 130 and statistical model(s) 120. Statistical model(s) 120 may be built using historical event data 125 for a given event type. The historical event data 125 for an event type may include data such as timing / acceleration characteristics and other data from prior events. In some embodiments, statistical model(s) 120 may incorporate at least the event duration and average speeds / accelerations of objects along the trajectory or path for the event. Specific event types can also be modeled to include scenes from a pre-event, start, features of interest, finish, post-event, etc.

[0037] In some embodiments, processing unit 110 may also include a tracking scheduler 160 configured to automatically track the movement of one or more objects engaged in an event by applying the 3D and statistical models 120 and 130. Together with a switcher 150, tracking scheduler 160 may be configured to schedule and coordinate the sequence and selection of image data feeds from one or more cameras (e.g., set of cameras 155) and / or other image capture devices. Additionally, tracking scheduler 160 may be further configured to provide a plurality of instructions (e.g., point of focus, pan, zoom, tilt, positions etc.) to one or more cameras (e.g., set of cameras 155) and / or other image capture devices. In FIG. 1A, one or more cameras 155 may beconfigured to provide image data feeds comprising images or videos during an event. Other image capture devices (not shown) may be provided to provide feeds of captured images or videos, depending on the requirements of a particular implementation. After the trajectory and sequence for an event is determined from the 3D and statistical models 120 and 130, tracking scheduler 160 may transmit and control the sequence via switcher 150 to capture image data for the event. Responsive to the sequence and selection from tracking scheduler 160, switcher 150 may be configured to manage and route the image data feeds from multiple sources, such as cameras 155, to various destinations. To this end, switcher 150 may be configured to activate and deactivate cameras 155 and / or other image capturing devices, provides instructions, and / or to select their respective feeds for routing and output, in accordance with the determined sequence provided by tracking scheduler 160. Accordingly, when the objects are engaged in an event, switcher 150 may generate a broadcast feed 180 from the selected sequence of image data feeds from one or more cameras 155 and / or other image capture devices. The broadcast feed 180 may be distributed across various remote devices 190, such as monitors or display devices.

[0038] In FIG. 1A, switcher 150 is depicted as a separate device external to processing unit 110. Consequently, in this configuration, the feeds from cameras 155 and / or other image capture devices bypass system 100. Alternatively, in other embodiments, switcher 150 may be integrated into system 100 as part of processing unit 110 or tracking scheduler 160. In such embodiments, processing unit 110 or tracking scheduler 160 may directly provide the controlling and routing functions of a switcher as described above, and one or more cameras 155 and other image capture devices may be connected directly to system 100. Thus, processing unit 110 or tracking scheduler 160 serves as the central point for managing the connected cameras and controlling the various feeds captured by the connected cameras and any other image capture devices. An exemplary embodiment of this configuration is shown in FIG. 1 B, where the set ofcameras 155 is directly connected to tracking scheduler 160. In this configuration, tracking scheduler 160 includes a switcher and is configured to activate and deactivate cameras 155 and / or other image capturing devices and / or to select their respective feeds for routing and output, in accordance with a determined sequence for a given event. Accordingly, when the objects are engaged in the event, tracking scheduler 160 of FIG. 1 B may generate a broadcast feed 180 from the selected sequence of image data feeds from one or more cameras 155 and / or other image capture devices. As with the embodiment of FIG. 1A, the broadcast feed 180 may be distributed across various remote devices 190, such as monitors or display devices.

[0039] As shown in FIGS. 1A and 1 B, an external controller 170 may be provided with a control interface 175. External controller 170 may communicate with processing unit 110 of system 100, as described above. Control interface 175 of external controller 170 may include hardware and / or software (e.g., a keyboard, mouse, joystick, display, graphical user interface, etc.) and be configured to allow an operator to interact with system 100 to perform operations via processing unit 110 or tracking scheduler 160, such as triggering the start of tracking an event or to make tracking or camera feed adjustments during the event. In addition, during the capture of an event, an operator or other user may supplement the automated tracking of the objects through control interface 175. By way of example, based on operator inputs from control interface 175, external controller 170 may send commands to tracking scheduler 160 or directly to cameras 155 (not shown) to optimize the framing of the objects engaged in the event and / or the capture of recorded images or video. In some embodiments, control interface 175 may be configured to enable an operator to change or modify the sequence of selection of feeds from cameras 155 and / or other image capturing devices via tracking scheduler 160.

[0040] In some embodiments, the automated tracking of objects may be refined based on feedback or adjustment signals from a computer-vision system or otherintelligent computer system. For example, as shown in the embodiment of FIG. 1 B, a computer-vision system 165 may be implemented as part of tracking scheduler 160. To implement computer-vision system 165, an image or video processor may be provided that is configured to process the feeds from cameras 155 and / or other image capturing devices and, as a result of analyzing the objects engaged in the event, provide feedback or adjustment signals to tracking scheduler 160 to optimize the framing of the objects and / or the capture of images or video. By way of example, computer-vision system 165 could assist with detecting the start of an event or provide speed adjustments for sequencing between cameras tracking the objects in the event. Depending on the requirements of the implementation, computer-vision system 165 could also be implemented as a separate component or device from tracking scheduler 160 and, therefore, separate from system 100 and processing unit 110. In such a configuration (not shown in FIGS. 1A-1 B), computer-vision system 165 could directly receive and process the feeds from cameras 155 and / or other image capturing devices and, as a result of processing the feeds, communicate with tracking scheduler 160 to provide feedback or adjustment signals for controlling the framing of objects or capture of images or video.

[0041] Furthermore, in some embodiments, processing unit 110 may include or be connected to a sensor module configured to receive, process, and analyze data from multiple sensors (not shown in FIGS. 1A-1 B). Sensor data may be used as a feedback mechanism that is provided to tracking scheduler 160 to assist with the automated tracking of objects during an event. Such sensors can be used to confirm, e.g., the start or progression of an event and / or the passing of objects by motion sensors along the trajectory or path for the event within the venue. In some embodiments, sensors may be used to confirm the movement of the objects with respect to the distances or points along the trajectory to define steps in the camera capture sequence and / or recording workflow.

[0042] In some embodiments, processing unit 1 10 may upload data to a server (e.g., to a cloud server or system on the Internet) via a communication link, network, or means (not shown in FIGS. 1A-1 B). For example, processing unit 110 may upload to a server the broadcast feed 180 generated from the image / video feed received from one or more image capture devices or cameras (e.g., set of cameras 155) or sensor data received from multiple sensors. Similarly, in some embodiments processing unit 110 may be configured to download data from a server via a communication link, network, or means. For instance, processing unit 110 may download historical event data 125 and map data 135 from a server. Once downloaded, this data may be subsequently processed by processing unit 110 or stored in a memory associated with processing unit 110.

[0043] The depiction of processing unit 110 as a single entity in FIGS. 1A-B should be understood as illustrative, as system 100 may comprise multiple processing units or devices. Each of these units may be tailored to execute specific tasks and interact seamlessly with the others. For instance, system 100 may encompass a variety of processing units: such as one for managing statistical model 120, another for 3D model 130, a third responsible for tracking scheduler 160, and potentially more processing units that are configured to communicate with another and other components such as switcher 150.

[0044] Consistent with embodiments of the present disclosure, computer- implemented systems and methods may be implemented for tracking objects within a venue using one or more statistical and 3D models. These models may be created using historical and map data, facilitating the generation of a trajectory and predicted movement along that trajectory by one or more objects engaged in a given event. Based on the determined trajectory and predicted movement of the objects, a sequence of image data capture may be defined and utilized for automatically tracking the objects within the venue. This arrangement operates akin to a feed-forward control system, where trackingaligns with predictions without the need for predominant manual control or complex processing of image / video feeds, thereby avoiding the drawbacks of conventional approaches, while providing a solution with enhanced accuracy and automation and at the same time greater reliability and efficiency. In certain embodiments, the accuracy of predicted trajectories may surpass a predetermined threshold (e.g., 90%, 95%, etc.). Nonetheless, it is to be appreciated that the systems and methods disclosed herein may still incorporate and receive feedback or refinement signals, as needed. While the exemplary embodiments described herein are presented in the context of tracking objects involved in an event within a venue, it is to be appreciated that the disclosed embodiments may be implemented in other settings or contexts, and within any environment or area of technology that requires the tracking of objects.

[0045] FIGS. 2A-2B are flowcharts showing exemplary processes 200a and 200b for tracking objects. More specifically, process 200a corresponds to an exemplary initialization process for tracking objects, while process 200b corresponds to an exemplary process for tracking objects during an event. In accordance with the disclosed embodiments, such processes may be executed by system 100 via the implementation of processing unit 110 and the other components described with reference to FIG. 1A or 1 B. In some embodiments, processes 200a and 200b may be performed in sequence, one after the other, or after a certain amount of time elapses between the completion of process 200a and the commencement of process 200b. In some embodiments, process 200b may be executed one or more times before process 200a is re-executed.

[0046] Reference is now made to FIG. 2A and process 200a. At step 202, a statistical model is provided based on historical event data for at least one event type. As part of step 202, one or more statistical models may be provided as input to processing unit 110 for processing and / or storage in a memory. In some embodiments, historical event data may be provided as input to processing unit 110, and one or more statisticalmodels may be generated, modified, updated, and / or stored by processing unit 110. Within the context of the present disclosure, an event includes any sort of race, tournament, or competition. Such an event may run along a fixed, predictable, and / or repetitive path, such as a track, field, or lane(s), between defined start and end positions / zones, and may involve one or more participants or other objects. An event relates to the occurrence, realization, or instantiation of an event type. Event types encompass the parameters and conditions under which participants compete in events, including the start and end positions / zones, the course layout, any obstacles or challenges along the way, and the specific rules governing the competition. Examples of event types include but are not limited to track and field races (athletics races), motoring races (rallying, Formula 1 (F1 ), Moto or other motorcycling events, etc.), cycling races, skating races, skateboarding races, rollerblade races, drone races, robot races, boat races (canoeing, kayaking, rowing, jet ski, paddle, etc.), skiing races (slalom, crosscountry skiing, biathlon, etc.), animal races (e.g., horses, greyhounds, sled dogs, camels, etc.), obstacle races, swimming competitions, sliding competitions (bobsleigh, sled, etc.) throwing disciplines (javelin, discus, shot put, hammer, etc.), precision disciplines (archery, shooting, curling etc.), and jumping disciplines (high jump, long jump, pole vault, etc.). It will be appreciated that this is a non-exhaustive list of examples and that embodiments of the present disclosure may be implemented for other event types.

[0047] Historical event data refers to any information or collection of information documenting past events, including positional data, metrics, and / or outcomes. Historical event data may include one or more of past participants’ / objects' trajectories, race distances, participants / objects' speeds (average and / or nominal), participants / objects' timings (average and / or nominal), accelerations (average and / or nominal), participant information, participant interactions, weather conditions, course conditions, equipment used, race strategy, race result and rankings, records, or trends. For example, in someembodiments, the historical event data for the at least one event type may include information related to object movements and interactions during past occurrences of the event type. Within the context of the present disclosure, an object encompasses any participant, entity, or other object actively engaged in the event. Objects may encompass athletes, teams, vehicles, boats, equipment, and any other relevant components integral to the event's execution and significance. For instance, if the event pertains to a greyhound race, the objects may represent the various greyhounds enlisted to compete in the race. Similarly, if the event centers around a car race, the objects may denote the diverse array of cars taking part in the competition. In another example, if the event corresponds to a javelin throw competition, the objects may correspond to the javelins thrown by the different competitors.

[0048] A statistical model refers to any representation or framework (algorithmic, mathematical, etc.) that describes the relationship between variables in a dataset. Statistical models can provide a systematic approach to understanding and analyzing data by quantifying patterns, relationships, and uncertainties. In some embodiments, statistical models are used to make predictions, draw inferences, and test hypotheses about the underlying processes generating the observed data. In the context of providing a statistical model based on historical event data for at least one event type, the statistical model may comprise a mathematical construct used to analyze and interpret past occurrences or events recorded or represented in the dataset (e.g., historical event data 125). In some embodiments, providing one or more statistical models involves formulating mathematical equations, probability distributions, and / or algorithms that capture the relationships and patterns present in the historical event data. Each statistical model may incorporate one or more variables representing any of the above-mentioned data to quantify and predict outcomes or trends associated with future events. For example, the outcome of a statistical model based on historical event data may include predicted oraverage trajectories, performance predictions, predicted race strategies, or any other relevant insights obtained / derived from historical event data. In some embodiments, the statistical model may be generated by employing supervised and unsupervised learning techniques for adaptation to various event types. For example, machine learning techniques, such as regression analysis, clustering, and pattern recognition, may be employed to provide a statistical model that captures the inherent relationships and patterns within the historical event data.

[0049] By analyzing historical data, a statistical model may be provided that offers detailed information about one or more specific segments of a fixed, predictable, and / or repetitive 2D / 3D path. For instance, a statistical model may reveal insights such as the average initial speed or acceleration of participants / objects as they commence a race or other event along the path. It may also shed light on the final average speed or acceleration achieved by participants as they approach the end of the race or other events. Furthermore, the model can elucidate the evolution of speed or acceleration at specific turns or bends along the path, providing valuable information on how participants / navigate these critical points. Such analyses contribute to a deeper understanding of event dynamics.

[0050] In some embodiments, an outcome of a statistical model may predict the course of an event with an accuracy surpassing a predetermined threshold. For example, the outcome of statistical model 120 based on historical data 125, may predict a trajectory with an accuracy above 90%, 95%, or any other suitable percentage. Accuracy evaluation may involve comparing metrics predicted by the statistical model against metrics measured once an event concludes. For instance, if the model predicts the duration of an event, the actual duration can be compared to the predicted duration post-event to gauge prediction accuracy.

[0051] Referring to FIGS. 1A and 1 B, processing unit 110 may be configured to provide statistical model 120 based on historical event data 125. Statistical model 120 may be stored in a memory included in processing unit 110 or separate from processing unit 110. Historical event data 125 may also be stored within processing unit 110 or outside processing 110. For example, historical event data 125 may be stored on a remote server or database, and processing unit 110 may download historical event data 125 to generate and provide statistical model 120. In some embodiments, processing unit 110 may be configured to generate statistical model 120. For example, processing unit 110 may employ machine learning algorithms (e.g., neural networks, regression, decision trees) in conjunction with historical event data 125 to create statistical model 120. Alternatively, in some other embodiments, a user may have created and stored statistical model 120 into processing unit 110 or in a memory accessible by processing unit 110. For example, a user such as a data scientist, researcher, or developer, may design and code for the statistical model. This may involve defining the structure of the model, specifying algorithms and techniques, setting parameters, and implementing any necessary pre-processing or post-processing steps. Once the statistical model is completed, it may be stored in processing unit 110 or a memory accessible by processing unit 110 and called by processing unit 110 whenever necessary. This storage could be in the form of files, libraries, or specific memory locations accessible by processing unit 110. In some embodiments, where historical event data 125 exists in a raw format, processing unit 110 may be configured to undertake data processing tasks. Processing raw data may encompass various data formatting, conversions, and / or other manipulations aimed at enhancing data readability and usefulness. For instance, processing unit 110 may be configured to perform data segmentation, i.e., categorizing data into meaningful groups or classes based on attributes such as event type, occurrence time, object / participant characteristics (e.g., age, gender), weather conditions, and geographic location (e.g.,altitude). Additionally, processing unit 110 may convert raw data into a standardized format suitable for analysis or modeling, perform data cleaning to rectify errors and inconsistencies, enhance data by integrating additional information or features, and / or integrate data from disparate sources into a cohesive dataset.

[0052] Referring again to FIG. 2A, at step 204, a 3D model is provided of a venue using map data. As part of step 204, one or more 3D models may be provided as input to processing unit 110 for processing and / or storage in a memory. In some embodiments, the 3D model may comprise complete map data for the venue or map data for select portions of the venue. In the context of this disclosure, a venue encompasses any location, building, or place designated to host one or more events of a same or different type. Venues may encompass a variety of settings, facilities, and environments tailored to accommodate specific event types, such as races or competitions. Examples of venues include stadiums, arenas (both indoor and outdoor), convention centers, racetracks, parks, bodies of water (lakes, rivers, ponds ... ), or any other suitable location, building, or place capable of hosting an event.

[0053] A 3D model refers to any digital representation of an object, venue, or environment. A 3D model for a venue may be created using, for example, three- dimensional computer graphics software, photogrammetry software, or CAD software. It simulates the spatial dimensions and physical characteristics of the real-world entity or scene, including length, width, height, shape, texture, and / or color. 3D models may be composed of geometric primitives such as points, vertices, edges, and faces, which are manipulated and arranged to form complex shapes and structures. 3D models may be stored using various data structures, such as a list of points and respective coordinates, mesh representations, or voxel grids, among others. 3D models may be visualized via a user interface. Such visualizations can be static or dynamic, allowing users to view, manipulate, and interact with the 3D model from different angles and perspectives. Withinthe context of the present disclosure, a 3D model of a venue may depict the architectural layout, spatial dimensions, features, design, and structural elements of the venue as well as its equipment. In some embodiments, the 3D model may include one or more event paths. The 3D model may also incorporate surrounding landscapes, infrastructure, and environmental features to provide a comprehensive visualization of the venue and its surroundings. For example, in some embodiments, the 3D model of a venue may comprise detailed representations of structural elements, seating arrangements, and environmental features within the venue.

[0054] As disclosed herein, a 3D model may be based on map data. Map data refers to any sort of information (e.g., geographic information, dimensional information, etc.) and spatial data that are used to create the model. Map data may be sourced from geographic information systems (GIS), GPS data, satellite imagery, site surveys (e.g., aerial surveys), engineering drawings (e.g., blueprints), LiDAR (Light Detection and Ranging) scans, photogrammetry with camera triangulation, and / or other mapping technologies. According to the nature of the map data source, various coordinate systems may be employed. For example, coordinate systems may include latitude and longitude for geospatial data, elevation coordinates, cartesian coordinates for local or projected mapping, or other specialized systems tailored to specific applications.

[0055] Referring to FIGS. 1A and 1 B, processing unit 110 may be configured to provide 3D model 130 based on map data 135. 3D model 130 may be stored in a memory included in processing unit 110 or separate from processing unit 110. Similarly, map data 135 may be stored within processing unit 110 or outside processing 110. For example, map data 135 and / or 3D model 130 may be stored on a remote server or database, and processing unit 110 may download map data 135 and / or 3D model 130 to provide 3D model 130. In some embodiments, processing unit 110 may be configured to generate 3D model 130. For example, processing unit 110 may employ three-dimensional graphicsprocessing software in conjunction with map data 135 to create 3D model 130. Alternatively, in some other embodiments, a user may have created / coded and stored 3D model 130 into processing unit 110 or in a memory accessible by processing unit 110. Processing unit 110 may therefore call 3D model 130 whenever necessary.

[0056] FIG. 3 illustrates an exemplary venue 300, i.e. , a stadium. Such a venue may host various events of different event types pertaining to athletics disciplines or competitions. For example, stadium 300 may host races of various lengths (100m, 200m, 400m, 1500m), hurdles, steeplechases, various throwing disciplines (javelin, discus, shot put, hammer), and different jumping disciplines (high jump, long jump, pole vault). Accordingly, a single 3D model of a venue may be used for different events of different types. Alternatively, it may be the case that a venue is only capable of hosting a single type of event such as a Formula 1 (F1 ) motorsports racetrack or a greyhound racetrack. Consistent with the present disclosure, different 3D models may be generated for different venues using the corresponding map data for each venue.

[0057] FIG. 4A illustrates an exemplary 3D model 400 of a racetrack 410 which could be situated within a stadium or other venue. Racetrack 410 features multiple lanes or corridors (430-1 through 430-4) and is marked with a starting / finishing line 420. It should be noted that the start and finish positions / areas do not necessarily have to be concurrent. For example, in a 400-meter race, participants start at different positions and finish at the same level. In another example, in a bobsleigh competition, the start and finish lines are separate. Racetrack 410 exhibits a grid pattern representing the different edges / faces used in the 3D modeling process. 3D model 400 illustrated in FIG. 4A is an exemplary embodiment and not necessarily exhaustive. For instance, if racetrack 410 is part of venue 300, a 3D model of venue 300 might also include the seating arrangements within the stadium or the surrounding environment of racetrack 410.

[0058] FIG. 4B illustrates another exemplary 3D model 450 of a different venue, more specifically a motorsports circuit 460 which may be obtained based on map data (e.g., GPS data, site survey, etc.). The starting / finishing line 490 of motorsports circuit 460 is represented by a black line. One or more different events could be hosted within circuit 460, each with different distances (e.g., number of laps), classes of vehicles, and racing leagues or federations. Consistent with the present disclosure, a statistical models may be created for each event type.

[0059] In certain embodiments, historical event data may encompass event data collected during past occurrences of events at various venues. For instance, a racetrack, such as a 400-meter racetrack, might exhibit minimal or no variation from one stadium to another. Consequently, historical event or race data may aggregate data from past races held at different stadiums. For example, outcome 510 may be the result of a statistical model based on historical event data originating from venue 300 and racetrack 410 but also from other similar racetracks at different venues. In some embodiments historical event data may encompass event data collected from past occurrences of events at a particular venue. For example, in the case of a F1 or other motorsports circuit such as that illustrated in FIG. 4B, only historical event data collected at that specific motorsports circuit and for its unique characteristics may be used for the statistical model. Consequently, the statistical model based on historical event data may be either agnostic or specific to a venue, depending on the specific requirements and characteristics of a venue, the statistical model may leverage historical event data from various sources or be tailored specifically to the characteristics of a particular venue.

[0060] In some embodiments, the venue may serve as a variable within the statistical model, and the statistical model's outcome can vary based on this variable. The model may possess the capability to discern relevant historical event data based on a venue indicator, thereby determining whether specific historical event data collected atdifferent venues should be aggregated and utilized for the model. This means that the statistical model could incorporate the venue as a factor influencing the outcome, allowing for the selection and inclusion of pertinent historical event data specific to each venue. By employing venue identification as a criterion, the model may adapt its analysis and decision-making process, potentially yielding more accurate predictions or insights tailored to the characteristics of individual venues.

[0061] In some embodiments, the outcome of the statistical model may be based on a specific set of historical event data. For example, the outcome of the statistical model may be calibrated using data from a limited number of recent events, such as the last five, ten, or twenty events hosted at a particular venue. In contrast, a 3D model is, by definition, specific to a venue. Even though certain event types may present similar characteristics across different venues, the 3D model remains inherently tied to the unique layout, dimensions, and features of a particular venue.

[0062] Optionally, at step 206, processing unit 110 may combine the 3D model and the statistical model. This step may result in the seamless integration of the 3D model and the statistical model to form a combined 3D and statistical model and unified framework for object tracking. This integration allows for the exploitation of both spatial and statistical information during the tracking process. Spatial coordinates from the 3D model may be utilized to correlate with statistical patterns, enabling a more accurate and context-aware determination of object trajectories, enhancing thereby tracking accuracy. This step of the process may include various operations. In some embodiments, the outcome of a 3D model may be used to determine a path (event path), i.e. , a route or course that one or more objects may follow from one point to another during an event. A defined path may correspond to the sequence of locations or positions occupied by the one or more objects as they move through space, accordingly a path may be created using mapped coordinates from the 3D model. Once a path is determined, the statisticalmodel may be applied, and its outcome may represent a trajectory based on the determined path. The trajectory may describe the motion of the one or more objects over time, as they move along the determined path. Accordingly, the 3D model may serve as a foundation for driving the statistical model. The spatial information provided by the 3D model may be utilized to guide and inform the statistical model. By correlating spatial coordinates from the 3D model with statistical patterns, the system may generate more accurate trajectories and predictions regarding the motion of the one or more objects being tracked.

[0063] For example, referring to FIG. 4A, 3D model 400 of racetrack 410 may be used to determine a path. Such a path may for example be located between the second 430-2 and the third corridor 430-3, as such an intermediate path following the central marking may be optimal for monitoring multiple runners within each corridor 430-1 through 430-4. Once this path is determined, the statistical model may be utilized to delineate a trajectory, indicating the pace at which the determined path will be traversed.

[0064] Alternatively, in some embodiments, the output of a 3D model may serve as the basis for establishing a preliminary path, which can then be refined using a statistical model. While the 3D model may offer initial insights into the path followed by objects during an event, it might lack specific details regarding how objects strategically navigate certain points such as bends or curves along the path. For instance, referring to FIG. 4B, although the 3D model 450 may outline the path that F1 cars or other vehicles are expected to follow, it may not capture the intricate maneuvers they make while negotiating bends or curves. In such cases, the preliminary path provided by 3D model 450 may be enhanced through the utilization of a statistical model, leveraging historical event data such as past race strategies. This historical data may help refine the raw path 460 inferred from the circuit layout. Moreover, the statistical model can determine the pace at which the circuit 460 will be traversed. A similar scenario arises in events related to throwingdisciplines such as javelin. While a 3D model of a stadium can indicate the positions of throwers, the landing area of a javelin, and the volume above the stadium field where the javelin could potentially be located after being thrown, it may not indicate the precise trajectory (parabola) followed by the javelin after being thrown. The use of historical event data via a statistical model may help to establish the average trajectory, including the height and spatial distribution of the javelin's path above the stadium field.

[0065] In yet other embodiments, a statistical model may be utilized to establish an estimation of the pace for the event along a designated path. This information may be used with a 3D model containing coordinates delineating the location of the path within the venue. For instance, statistical parameters like speed and distance traveled can be leveraged to compute a sequential progression along the path over discrete time intervals, i.e., assessing the hypothetical distance covered by a tracked object as a function of time. This calculated pace along the path may then be utilized with the 3D model, which encompasses spatial coordinates representing the location of the path within the venue. By doing so, the trajectory at the venue may be determined based on the established pace. Additionally, in certain embodiments, considerations for characteristics such as the topography of the venue provided by the 3D model are factored in (e.g., by calculating how they can increase or decrease the pace) to regulate and update the trajectory as necessary.

[0066] Referring to FIG. 1 B, processing unit 110 may combine statistical model 120 based on historical event data 125 to 3D model 130 generated with map data 135, to generate a combined model 140. Combined model 140 may then be stored in a memory within processing unit 110 or outside processing unit 110. Combined model 140 may also be uploaded to a server. Alternatively, as shown in FIG. 1A, statistical model 120 and 3D model 130 may not be combined but utilized individually and together for, among other things, determining a path for an event and a trajectory along that path to predictmovements of the objects engaged in the event. It will be appreciated that exemplary methods and features disclosed herein, including in FIGS. 2A and 2B, may be implemented using the system of FIG. 1 A or 1 B and irrespective of how the statistical and 3D models are combined or applied.

[0067] FIG. 5 depicts an exemplary outcome 500 derived from the combination of 3D model for a venue housing racetrack 410 and a statistical model based on historical event data associated with events occurring on racetrack 410, specifically race events of a certain distance. Alternatively, outcome 500 may be obtained by using the output of 3D model 410 and the output a statistical model based on historical event data associated with events occurring on racetrack 410 independently, or with the output of one model serving as input for the other model. The outcome displayed here represents an average trajectory 510 along the racetrack 410 created using mapped coordinates. This trajectory is divided into multiple segments delineated by nodes (discrete 3D points in space) represented by black crosses. Each node corresponds to a specific point along the racetrack, and the distance covered is indicated in its vicinity. Each node may be placed into a data structure indexed by distance (the distance value may be created by calculating the distance between two nodes). The trajectory is coded according to a grayscale map 530, indicating the speed or average speed on each particular segment. Similarly, each speed or average speed may be placed into a data structure. Additionally, outcome 500 includes supplementary information 520 such as the name of the venue, prevailing weather conditions during the average trajectory 510, as well as the average and maximum speed and distance covered. In some embodiments, because of the finite resolution of the 3D model, one or more nodes or positions along the determined trajectory may be obtained by interpolation. In some embodiment, one or more distances covered may be determined in connection with the predicted trajectory. These distances may represent the covered distances by one or more hypothetical objects along thepredicted trajectory. For instance, in the context of a race, a first covered distance (measured against time) might be linked with the anticipated distance covered by a racer in pole position, thus indicating a faster pace compared to other racers. Conversely, the second covered distance could signify the anticipated distance covered by the last racer. By having these two distance metrics, it becomes feasible to generate more precise instructions for transmission to a set of cameras, ensuring effective tracking of all racers throughout the race.

[0068] In some embodiments, the outcome of the statistical model or combined model may vary based on one or more initial conditions. For instance, different weather conditions could influence the outcome; for example, average speed may decrease during hot or rainy weather. Furthermore, the expected average speed may differ depending on whether the event corresponds to a qualifying race or a final race, as runners tend to perform better and run faster during final races. In some embodiments, the statistical model may incorporate a trained model derived from historical event data and designed to adjust the model's outcome based on initial conditions. Additionally, or alternatively, in some embodiments, the statistical model may be trained using a combination of supervised and unsupervised learning techniques to adapt to various event types and venue configurations.

[0069] As indicated above, metrics predicted by the statistical model may be compared to metrics gathered after an event concludes. For example, referring to FIG. 5, various metrics like predicted average speed, maximum speed, race duration, average speed along specific segments, or other pertinent metrics may undergo comparison with actual metrics determined post-event. In some embodiments, the accuracy of the outcome of the statistical model may be above a predetermined threshold (e.g., 90%, 95%, etc.). If the determined accuracy falls below the predetermined threshold, it suggests that one or more unforeseen events may have occurred during the event.Additionally, it may indicate that the statistical model requires updating with more recent historical data. This adjustment ensures that the model reflects the latest trends and patterns, thereby enhancing its ability to provide accurate predictions and insights into event outcomes.

[0070] While outcome 500 of the statistical model (e.g., statistical model 120) is presented here as an illustration and that may be displayed via a user interface associated with system 100, it is to be appreciated that the outcome of a statistical model or combined model can manifest in various forms and utilize diverse formats. These may include spreadsheets, tables, files (e.g., XML files, CSV files, text files, JSON files), graphs, charts, visualizations, reports, dashboards, raw data, or any other suitable form / format. For instance, details pertaining to the average trajectory 510 could be incorporated into a spreadsheet format.

[0071] Referring to FIG. 2A, at step 208, processing unit 110 may map into the 3D model or combined 3D and statistical model a location for each of a set of cameras installed at the venue. A set of cameras (e.g., set of cameras 155) installed at the venue (e.g., venue 300) included any sort of device capable of capturing images and transmitting an image feed. The mapping process may involve integrating the physical locations of cameras installed within a venue, such as venue 300, into the 3D model or combined 3D and statistical model. This process may entail different steps such as identifying the specific positions of each camera within the venue's space and translating them into coordinates within the 3D or combined model's framework, which may include dimensions in 3D space and statistical parameters. By incorporating the mapped coordinates of the cameras into the model, the system may establish a spatial relationship between the cameras and the venue environment. Each camera may be assigned a unique identifier within the model for tracking and analysis purposes. Additional steps of calibration and validation steps may be implemented to ensure and improve, if necessary, the accuracyand reliability of the mapping, verifying the correspondence between physical locations and mapped coordinates.

[0072] In some embodiments, processing unit 110 may be further configured to map a location for each of one or more additional devices apart from cameras. Examples of such additional devices may encompass but are not restricted to spotlights, microphones, sensors, or any other equipment capable of capturing data relevant to an event or enhancing the quality of the captured or recorded data.

[0073] Once the initialization process for tracking objects, denoted as process 200a, is completed, processing unit 110 may initiate the execution of process 200b of FIG. 2B. As previously noted, a single initialization process may suffice for multiple tracking instances of objects within a same venue across consecutive events. For instance, in the scenario where multiple similar races (of a same event type, i.e. , same distance, same style etc.) unfold on racetrack 410, the 3D and statistical models (e.g., models 120 and 130) or the combined 3D and statistical model (e.g., combined model 140), forged during the initialization, may persist without alteration. This continuity ensures efficiency and avoids redundancy in recalibrating models for each successive event, which may be only a few seconds or minutes apart. Alternatively, if two different types of events are scheduled consecutively, such as races of varying lengths or different styles (e.g., sprint and hurdles), the initialization process might require repetition to accommodate the transition between event types. In yet another scenario, multiple models may already be initialized and stored in a database waiting to be called for particular event types. Referring to racetrack 410 presented in FIG. 4A, different events of different types (e.g., races of different length) may occur consecutively at this venue, and different models (e.g., different statistical models) may be pre-initialized and stored within a database, poised to be deployed for specific event types. For example, a first 400m race may occur, and models tailored specifically for this race type may be promptly invoked. Subsequently,if an 800m race is scheduled, models optimized for this particular event type may be called into action without delay. This pattern continues seamlessly, even if the event types shift rapidly. For example, if just a second after the completion of the second race, a third 100m race is scheduled the corresponding model may be called. All pertinent data necessary for analysis and decision-making may be stored across one or more permanent databases, ensuring instantaneous access. Consequently, system 100 may seamlessly transition from one event type to another, obviating the need to retrieve or store data beyond the latest iteration of the statistical model or execute the initialization process 200a.

[0074] Referring to FIG. 2B, at step 210, processing unit 110 may identify a commencement of a new event at the venue. The term "commencement" generally refers to the initiation of an event, whether this is the start of a race or the performance of new participants / objects (e.g., in a pole vault competition, the jump of each participant may be considered as a separate event). Moreover, "commencement" may also encompass the moments preceding the official beginning of the event. For example, it may refer to the time when participants are lining up, and preparing themselves before the event kicks off. Processing unit 110 may identify the commencement of an event through various means. One method may involve receiving one or more trigger signals. For instance, in certain embodiments, the processing unit may be linked to an event timer designed to supervise and manage the initiation or unreeling of the event. This event timer might be under the control of an official jury or referee. Alternatively, processing unit 110 could receive signals from an external controller, such as external controller 170 depicted in FIG. 1A or 1 B. If this external controller is connected to a user interface (e.g., via control interface 175), an operator might transmit a trigger signal to indicate the start of a new event. Additionally, or alternatively, processing unit 110 might be connected to one or more devices that are either involved in or configured to detect the commencement of the event.For example, it could be linked to a gridline light for a car race, an alarm set to ring at the event's outset or a starting gun, or various sensors such as motion sensors in animal boxes that detect when the boxes open and the animals begin racing. As a still further example, RFID or NFC chips or sensors may be used to record when one or more objects / participants engaged in an event pass a landmark. Moreover, in some embodiments, a computer-vision system (e.g., computer-vision system 165 or another intelligent system in communication with processing unit 110) may be utilized to determine when the event has started or is about to start. Such a system can analyze visual cues and patterns to identify the commencement of the event, adding another layer of detection capability to the overall process.

[0075] At step 212, processing unit 110 may determine a trajectory for one or more objects for the new event based on the 3D and statistical models or combined 3D and statistical model. As mentioned above, an object refers to any entity or participant actively engaged in an event. For example, the one or more objects may be runners engaged in a race or other competition. Through the integration of statistical parameters such as speed or accelerations and the utilization of a 3D coordinates processing unit, processing unit 110 can effectively ascertain an anticipated trajectory for one or more objects involved in the event. By harnessing statistical metrics alongside advanced processing capabilities in 3D coordinates, processing unit 110 may be empowered to predict the likely path and trajectory of objects engaged within the event with enhanced accuracy and reliability. This approach enables the tracking system to dynamically predict the movement patterns and behaviors of the objects, thereby projecting their trajectories with a high degree of precision and foresight.

[0076] It is to be appreciated from the present disclosure that the determined trajectory may not strictly correspond to the exact trajectory followed by the one or more objects for every event, but rather facilitates the tracking of the one or more objects orclosely approximates their actual trajectories. For instance, referring to FIG. 4A, where each runner remains in their assigned corridor or lane, processing unit 110 may not determine four distinct trajectories but rather a trajectory enabling the simultaneous tracking of all runners, such as one following the central marking. In events where participants follow one another (e.g., a motorsports race), the determined trajectory may approximate or closely correspond to the actual trajectories of the one or more objects. However, during the event, specific participants may overtake others, momentarily deviating from the determined trajectory. The predicted trajectory accommodates such events, enabling continuous tracking of all objects throughout the event, even during maneuvers like overtaking. In certain scenarios, the determined trajectory may align with the trajectory of a group of objects or participants, such as the trajectory of a pack of dogs during a greyhound race or a peloton of cyclists during a bike race.

[0077] FIG. 6 illustrates an exemplary representation of an outcome 600 of the 3D and statical models or combined 3D and statistical model established for racetrack 410 in venue 300 within 3D model 400. Outcome 600 includes a determined trajectory 610, following the central marking of racetrack 410 that encompasses various statistical information such as speed, accelerations, or distances. FIG. 6 further illustrates various participants 660 represented as cylinders on the starting / fin ish ing line 420. In accordance with the disclosed embodiments, trajectory 610 corresponds to a trajectory that enables tracking of the one or more objects 660. In some embodiments, once established and initialized (such as through steps 202 to 208 shown in FIG. 2A), the statistical model or the combined 3D and statistical model (e.g., combined model 140) may be supplied with one or more initial conditions prior to the commencement of a new event. Consequently, the determined trajectory of the one or more objects may vary between successive events. For instance, if there is a sudden alteration in weather conditions between two consecutive events, the initial conditions might account for this change, thereby affectingthe determined trajectory before the onset of a new event under the updated weather conditions. This process of fine-tuning or tweaking the combined model may significantly enhance the accuracy of the determined trajectory without necessitating a complete overhaul of the model itself. This fine-tuning involves making subtle adjustments to specific parameters or aspects of the model, thereby refining its predictive capabilities in response to evolving conditions or new data inputs. This approach not only optimizes computational resources but also ensures that the model remains agile and responsive to real-world dynamics by being updated before the beginning of any new events, thereby enhancing its overall effectiveness in trajectory prediction tasks. One aim of this fine- tuning may be to ensure that the accuracy of the model remains above a certain predetermined threshold (e.g., 90%, 95%, etc.).

[0078] In some embodiments, an initial trajectory may be established prior to the start of a new event, serving as a projected path for the event at the given venue. This initial trajectory could be determined independently of the latest event conditions. For instance, factors like weather conditions or the specific object / participant involved may not be taken into consideration during the determination of this initial trajectory. Following its determination, the initial trajectory might undergo adjustment or refinement by incorporating one or more initial conditions associated with a new event before the commencement of the new event. As further described below, this may be done to enhance predictive capabilities and refine trajectory accuracy. Alternatively, if initial conditions are absent or if discrepancies among various predictions are negligible, the initially determined trajectory may be adopted as the definitive trajectory for the new event.

[0079] At step 214, processing unit 110 may determine a sequence of image data capture for the set of cameras installed at the venue based on the determined trajectory. Referring to FIGS. 1A and 1 B, this sequence may be determined by tracking scheduler160 using the 3D and statistical models (e.g., models 120 and 130) or the combined 3D and statistical model (e.g., model 140) for activating / deactivating and selecting the feeds from the set of cameras 155. In the configuration illustrated in FIG. 1A, the image / video feed of set of cameras 155 bypass system 100 via switcher 150. Alternatively, as mentioned earlier, processing unit 110 or tracking scheduler 160 may take on the role of the switcher 150 and directly send activation / deactivation and feed sequencing instructions to the set of cameras 155, as shown in FIG. 1 B. In this configuration, processing unit 110 may also receive image data of the one or more objects captured by the set of cameras in the sequence determined based on the trajectory.

[0080] The determined sequence is utilized to automatically track the one or more objects throughout the event. In this context, tracking entails the continuous monitoring of the one or more objects during the event, ensuring that image data representing their presence at the venue is captured at any given time. However, it is to be appreciated that the presence of objects or the unfolding of an event is not a prerequisite for any of the steps of processes 200a-b presented herein. The sequence of image data capture may be determined and transmitted to the installed cameras regardless of whether objects are present at the venue or an event is unfolding. This sequence of image data capture corresponds to a series of commands issued to one or more cameras and does not necessarily require the presence of objects at any point. For instance, during a testing phase, the sequence for image data capture can be determined and transmitted to one or more cameras without any event occurring or objects being present at the venue.

[0081] In situations where the locations of one or more additional devices have been integrated into the combined model (similar to step 208 of FIG. 2A for cameras), processing unit 110 may also be programmed to establish a sequence for activating and deactivating these devices. For instance, if the positions of a set of spotlights installed at the venue have been mapped, processing unit 110 can determine the order for turningthe set of spotlights on or off. This setup aims to enhance the quality of the image data captured by the installed cameras at the venue by ensuring adequate illumination for the objects during the event. By orchestrating the activation and deactivation of spotlights or other devices in a systematic sequence, the system optimizes the lighting conditions for better visibility and clearer imaging of the objects being monitored.

[0082] In some embodiments, the set of cameras may include a combination of fixed and movable cameras. A fixed camera refers to a stationary camera that is mounted in a specific position and angle to capture a particular view of the event. These cameras can be installed in strategic locations around the venue to provide consistent coverage of key areas in the venue such as a track or field. Fixed cameras offer stable and reliable footage of the action from a predetermined perspective. On the other hand, a movable camera, also known as a mobile camera or a roaming camera, refers to a type of camera that can be moved and repositioned during an event to capture different perspectives and angles of the action. Unlike fixed cameras, which remain stationary throughout the event, movable cameras offer flexibility and agility in capturing dynamic moments from various vantage points.

[0083] Examples of cameras that may be used to cover an event include but are not limited to:• PTZ (Pan-Tilt-Zoom) Cameras: PTZ cameras are versatile camera systems capable of panning (horizontal movement), tilting (vertical movement), and zooming to capture dynamic scenes and follow the action during an event in real-time. These cameras may be remotely controlled to adjust their position, angle, and zoom level to focus on specific areas or objects / participants during the event.• Spider Cam: A Spider Cam is a specialized camera system suspended by cables above the venue, allowing for dynamic aerial shots and unique perspectivesof the action. It consists of a camera attached to a motorized carriage that moves along a network of cables, providing smooth and sweeping shots from various angles.• Camera Cranes: Camera cranes are equipped with remote-controlled camera systems mounted on telescopic arms. They may be elevated to capture high-angle shots or lowered to ground level for dramatic perspectives. Camera cranes are often used in events to capture sweeping shots of the action from different heights and angles.• Robotic Cameras: Robotic cameras are automated camera systems controlled remotely by operators or computer software. They may be programmed to track or follow objects / participants or capture specific areas of interest in the venue. Robotic cameras offer precision and consistency in capturing the action, making them ideal for live broadcasts and replay analysis.• Rail Cam: A camera on rail, also known as a rail cam or rail-mounted camera, refers to a camera system that is mounted on a rail or track for controlled movement along a predefined path. This setup allows for smooth and consistent tracking shots of the action from various angles and perspectives.• Drone-mounted camera: A drone-mounted camera refers to a camera that is attached to an unmanned aerial vehicle (UAV), or drone. These cameras are specifically designed to be lightweight and compact to be carried by drones while still maintaining high-quality imaging capabilities. Drone-mounted cameras offer perspectives and angles that are not easily achievable with traditional ground- based cameras, making them valuable tools for tracking scenes. Drone-mounted cameras may be used to closely follow one or more objects throughout an entire event (e.g., from start to finish line).• Tethered balloon camera: A tethered balloon camera refers to a type of aerial imaging system that consists of a camera suspended beneath a tethered balloon. The balloon is anchored to the ground via a tether, which provides stability and control over the balloon's altitude. The camera, typically mounted on a stabilized platform, captures images or videos from an elevated perspective as the balloon ascends.• Participant-mounted camera: A participant-mounted camera refers to a camera that is attached to a participant or object (e.g., car, helmet etc.) actively engaged in an event. Unlike stationary or other movable cameras, participantmounted cameras move in tandem with the individual or object, providing a unique point of view that captures the action from the participant's perspective.• High-Speed Cameras: High-speed cameras are equipped with highspeed imaging capabilities, allowing them to capture rapid sequences of images with precise timing. Such cameras may be used in various competitive sports, particularly in racing events, to determine the precise finishing order of participants or for false-start detection. Such cameras may be activated by sensors or triggers placed at the starting line and / or the finish line. These triggers are often linked to the cameras and are activated when participants cross these points. Those cameras may also be synchronized with the starting gun or an electronic starting signal. When the race begins, the camera detects the sound or signal and starts recording.

[0084] In some embodiments, the set of cameras may include infrared or low-light cameras. These specialized cameras are designed to capture images in conditions where traditional visible light cameras struggle to produce clear images, such as low-light environments or scenarios where illumination is limited. By utilizing infrared or low-light cameras, systems can extend their capabilities beyond conventional visible-light imaging,providing enhanced vision in challenging environments where traditional cameras may fail to produce usable images. Such cameras may notably facilitate tracking in low- visibility conditions.

[0085] FIG. 6 illustrates a set of cameras whose locations have been mapped into the combined 3D and statistical model (e.g., via step 208 of process 200a). The set includes a mixture of stationary and mobile cameras strategically positioned around racetrack 410. In detail, the set of cameras consists of fixed cameras, including a highspeed camera 620 positioned near the starting / finishing line 420, and cameras 650-1 through 650-4 located at the four comers of racetrack 410. Additionally, the arrangement includes movable cameras such as rail cams 630-1 and 630-2 flanking the straight segments of racetrack 410, as well as rotating cams 640-1 and 640-2 positioned at the center of the curved sections of racetrack 410. Given this configuration, and in consideration of determined trajectory 610, processing unit 110 may determine a sequence of image data capture for the set of cameras as follows: Initially, the commencement of the event is captured by high-speed camera 620. Subsequently, rail cam 630-1 assumes the recording, followed by camera 650-1 , rotating cam 640-1 , camera 650-2, rail cam 630-2, camera 650-3, rotating cam 640-2, camera 650-4, then back to rail cam 630-1 , and ultimately returning to high-speed camera 620 for the photo finish. It is to be appreciated that two or more cameras may record simultaneously or during overlapping time intervals. The previous sequence provided is merely indicative of a potential order of image capture, but in practice, the actual recording may occur concurrently or with overlap among the different cameras. In some embodiments, a sequence may be repeated if the new event involves multiple laps or if the determined trajectory loops over itself multiple times. Moreover, in some embodiments, if across multiple consecutive events, the determined trajectory is maintained or does not differsubstantially, the sequence may be maintained. For example, the sequence may be identical over multiple identical races.

[0086] It is to be appreciated that the determined sequence may encompass more than just the order of activation and deactivation or selection of camera feeds. It may also involve specifying settings to be applied to a particular camera at a given moment (such as zoom, focus, etc.) and / or instructions to adjust the positions of movable cameras (like pan, tilt, etc.) at specific time intervals.

[0087] In some embodiments, the sequence of image data capture may be organized based on the covered distance during the event. For instance, referring to FIG. 6, each node along the determined trajectory 610 (depicted as black crosses) may be associated with triggers, such as instructions or metadata, for the array of cameras installed at venue 300 (such as cameras 620, 630-1 , 630-2, 640-1 , 640-2, 650-1 , 650-2, 650-3, and 650-4). For instance, a node positioned around the first bend, situated at a specific distance from the starting line 420 of racetrack 410, could be linked to instructions or metadata for camera feed switching from camera 630-1 to camera 640-1 . Moreover, once the sequence of image data capture is determined, it can be applied in real-time to (e.g., via switcher 150) to select the feeds among the cameras as the event progresses, i.e. , as the covered distance increases. Alternatively, one or more instructions within the sequence of image data capture might be applied ahead to accommodate for latencies. For example, such a configuration may allow for accommodating for any latencies present in the camera systems (such as due to mechanics, focus adjustments, etc.), ensuring that the cameras within the set are prepared precisely when needed. In another scenario, the entire sequence of image data capture could be transmitted or applied to the set of cameras prior to the commencement of the event. In this setup, all cameras within the set are pre-instructed regarding their operation throughout the event.

[0088] In some embodiments, the sequence of acquisition / image data capture may be further determined by including optimization for energy efficiency. This approach may help in sustaining tracking capabilities during prolonged events while minimizing power consumption. The tracking system may intelligently control camera movements to prioritize key areas of interest and reduce unnecessary panning or tilting. Additionally, it may dynamically adjust the frequency of image or video data capture based on event requirements and object movement patterns, conserving power during periods of low activity. Selective data transmission strategies may prioritize the transmission of critical data while storing less urgent information locally for later analysis, reducing energy overhead. Low-power operation modes and predictive energy management may further optimize energy usage by leveraging sleep modes, power gating, and proactive adjustment based on historical data and event patterns. Overall, these strategies collectively enhance energy efficiency without compromising tracking performance, ensuring reliable operation throughout extended events. For example, as depicted in FIG. 6, following the determined trajectory 610, once objects 660 have passed the initial bend, the sequence of image or video data capture may involve deactivating cameras 640-1 , 650-1 , and 650-2. This deactivation may persist until the commencement of a new event or the start of a new lap.

[0089] In some embodiments, the orientation and / or position of the movable cameras may be controlled and adjusted according to the determined trajectory. For example, referring to FIG. 6, the positions and orientations of movables cameras 630-1 and 630-2 as well as cameras 640-1 and 640-2 may be controlled and adjusted according to determined trajectory 610. Additionally, or alternatively, in some embodiments, the movable cameras may have preconfigured camera paths and the preconfigured camera paths may be adjusted according to the determined trajectory. For example, each of rotating cams 640-1 and 640-2 may have a preconfigured path involving a 180° sweep tostick to the one or more objects as they navigate curved portions of racetrack 410. The parameters governing the execution of this sweep, including the starting point in time and angular velocity, may be modified / adjusted based on determined trajectory 610. Similarly, rails cams 630-1 and 630-2 may be programmed to travel their respective rails in either direction. However, aspects such as the speed profile or starting point of these paths may be determined or adjusted on the basis of determined trajectory 610.

[0090] In some embodiments, the trajectory of the one or more objects may be determined based on optimizing for camera coverage, minimizing blind spots, and / or maximizing the quality of captured image data. The trajectory planning may ensure that objects pass through areas covered by cameras to enable sufficient visual observations from different angles and perspectives. Blind spots, where cameras have limited or no coverage, may be minimized by positioning cameras to provide continuous and overlapping coverage of the monitored area. The trajectory planning may also consider factors affecting image quality, such as lighting conditions, camera settings, and environmental obstacles. By optimizing the trajectory, the system aims to enhance the clarity, resolution, and reliability of captured images. If the determined trajectory results in poor camera coverage, blind spots, or zones with low-quality data, an alert signal may be raised to prompt adjustments to camera parameters like position, orientation, focus, angles, or exposure settings. In cases where a fixed camera's physical position needs alteration, an operator must directly modify the camera's position at the venue. The new position may be then mapped into the combined 3D and statistical model to ensure an accurate representation of camera locations. It will be appreciated from this disclosure that whenever there are changes in the actual physical locations of the cameras, or if a camera is added to or removed from a camera set (e.g., due to malfunction), the mapping of the camera set into the 3D model or the combined 3D and statistical model (e.g., as conducted in step 208 illustrated in FIG. 2A) may be updated or adjusted. Consequently,the sequence derived (e.g., at step 214 illustrated in FIG. 2B) based on the determined trajectory may be adjusted to accommodate these changes.

[0091] In some embodiments, the set of cameras capturing image data of the one or more objects for a new event may constitute a subset of the cameras positioned at the venue, whose locations have been mapped into the 3D model or the combined 3D and statistical model. The process of selecting such a subset of cameras could depend on one or more factors, such as the specific event type and / or determined trajectory, with a particular subset of cameras being utilized based on the nature of the event and / or characteristics of the trajectory (such as the position, location, and / or length of the trajectory in the venue). For instance, in the scenario depicted in FIG. 6, where the new event encompasses a full turn of racetrack 410, all the illustrated cameras may be employed for capturing relevant data. However, if a different event type occurs, such as a shorter race covering a trajectory along only a portion of racetrack 410, the set of cameras utilized may be selected for that event and vary accordingly.

[0092] In some embodiments, processing unit 110 may receive from the set of cameras installed at the venue a signal indicating their status (e.g., positions, state etc.). This signal may be received either directly from the cameras or via another device (e.g., switcher 150). Processing unit 110, through tracking scheduler 160, may then analyze this signal to control and verify the image capture process. Additionally, this signal may serve as feedback for processing unit 110 to potentially adjust the determined trajectory during the event. For instance, if one of the cameras installed at the venue reports malfunctioning, field of view obstruction, or if it is not positioned as expected, tracking scheduler 160 may dynamically alter the sequence to ensure efficient tracking.

[0093] FIG. 7 illustrates the camera coverage of the set of cameras positioned at venue 300. Each camera's field of view is depicted as a grey shaded area, including 720 for the high-speed camera 620, 730-1 for rail cam 630-1 , 750-1 for camera 650-1 , 740-1for rotating cam 640-1 , 750-2 for camera 650-2, 730-2 for rail cam 630-2, 750-3 for camera 650-3, 740-2 for rotating cam 640-2, and 750-2 for camera 650-2. These fields of view may be either fixed or dynamic based on the camera type. In the illustration, certain fields of view may overlap with each other. The determined trajectory 610 is entirely covered by the camera coverage, with no blind spots. This comprehensive coverage may ensure that the trajectory of objects is effectively monitored and observed from multiple perspectives, enhancing the system's ability to track the one or more objects within the venue during the new event. It is to be appreciated that for specific event types and venues, a single camera may provide a wide enough angle to encompass and capture the one or more objects throughout the entire event, effectively tracking them all by itself. For instance, a wide-angle camera positioned on the roof of stadium 300 (not shown in FIGS. 6 and 7) might be capable of capturing the entire racetrack 410 and tracking participants 660 during the new event. However, even in such configurations, additional cameras (such as the one illustrated in FIGS. 6 and 7) may be employed to track the one or more objects, offering diverse perspectives. In certain venues, like the motorsport circuit 460 illustrated in FIG. 4B, a single camera might not provide comprehensive coverage, in such cases the use of multiple cameras for effective tracking may be required.

[0094] In some embodiments, the movement of the one or more objects along the determined trajectory may be analyzed via image data captured by the set of cameras installed at the venue. This analysis process be in real-time or during post-processing to determine the one or more objects' positions, velocities, accelerations, and / or other relevant parameters over time. The analysis may involve a set of steps such as• Detection: The presence of the one or more objects may be detected within the cameras’ fields of view. This can be done using various techniques such as motion detection, object recognition, or feature extraction. For example, in someembodiments, the image data captured by the set of cameras may be processed in real-time using machine learning algorithms to identify and classify objects. Such machine learning algorithm (e.g., Open CV) may be used to identify and classify various objects, such as racing vehicles and / or animals based on specific features (e.g., shape, color, size, or racing number)• Localization: Once the one or more objects are detected, their spatial coordinates may be determined within the camera's frame of reference or the combined 3D and statistical model. This may involve identifying the object's position, orientation, and size relative to the cameras’ viewpoints.• Tracking: The one or more objects’ positions / movement may be continually monitored as they traverse the venue / monitored area. Tracking algorithms may analyze successive frames of video or images to estimate the one or more objects' trajectories and / or predict their future path. Alternatively, the path of the objects may be determined based on the established trajectory. For instance, adjustments to the positions of the objects may be made if one or more participants are found to be ahead or behind relative to the projected trajectory. Tracking may be performed by using the image data of one or more cameras (e.g., two cameras offering different views).

[0095] Referring to FIG. 1 B and FIG. 6, the movement of objects 660 along determined trajectory 610 may be analyzed by computer-vision system 165 using the image feeds of the set of cameras 155 and / or other image capture devices. In certain embodiments, within the set of cameras, only a specific subset may be designated for movement analysis purposes. In such cases, only the camera feeds from selected cameras may be utilized and subsequently analyzed for determining the movement of the objects of interest. Meanwhile, the feeds from the remaining cameras within the set may serve other functions, such as broadcast feeds or other non-related activities.Alternatively, the feeds from other image capture devices may be processed by computervision system 165 instead of any of the feeds from cameras 155.

[0096] Furthermore, in some embodiments, processing unit 110 may be configured to adapt the statistical model in real-time based on live data obtained during the new event. Real-time adaptation based on live data in the context of camera tracking the trajectories of one or more objects / participants in a race or other event refers to the instantaneous processing, analysis, and incorporation in the combined model of data as it is collected by the set of camera systems. In this context, real-time means that the set of cameras is continuously capturing video footage and the system processes it immediately to track the movements and trajectories of objects / participants engaged in an event without significant delay. Data obtained this way may be immediately utilized to update / adapt a statistical model and / or its outcome. This includes dynamically adapting the statistical model to changes in event conditions, object / participant speeds, and unexpected behavior on the fixed 3D path of the event. The instantaneous processing allows for timely feedback, accurate monitoring, and rapid decision-making during the event. For example, the initial speed of a participant at a particular event may be used to update the estimated trajectory. In other words, based on the one or more objects' past movement patterns and current velocity, the combined 3D and statistical model may predict their future positions and behaviors and update the determined trajectory. This helps anticipate the object's path and ensure proactive camera adjustments to maintain optimal tracking. For example, providing feedback to the cameras to adjust their orientation, positions, zoom level, focus, frame rates, or other parameters to keep the object within view and maintain accurate tracking.

[0097] Additionally, or alternatively, in some embodiments, a trajectory of the one or more objects may be determined by dynamically adjusting the statistical model based on real-time sensor data. This real-time sensor data may originate from a sensor moduleintegrated within processing unit 110 or from an external sensor module connected to processing unit 110. Accordingly, processing unit 110 may be further configured to adapt the statistical model in real-time based on real-time sensor data. Real-time sensor data, in the context of sensors attached to objects / participants in a race or other event, refers to the immediate and continuous collection, processing, and transmission of information regarding various aspects of the object’s / participant's performance and / or physiological parameters during the race. These sensors may be placed along the trajectory and / or attached to different parts of the participants' bodies, such as GPS trackers, accelerometers, motion sensors, or heart rate monitors. As the race progresses, these sensors gather data in real-time, providing insights into factors like speed, distance covered, acceleration, biomechanical metrics (stride length, cadence), or heart rate. It is to be appreciated that such sensors may also be attached to objects per se such as cars, javelins, bikes, sleds, hammers, or any other relevant objects in movement during an event. Similar types of information apart from biological and biomechanical metrics may be collected. The term "real-time" emphasizes that the data is processed and made available instantly, without any significant delay. This immediate feedback allows adjustments to the statistical model and gain valuable insights as the event unfolds. For example, the instantaneous acceleration of a participant at a particular event may be used to update the estimated trajectory of such a participant during the event. Real-time sensor data may also be used by processing unit 110 for live tracking, or performance analysis. Referring to FIGS. 1A and 1 B, each of these sensors may communicate with processing unit 110 via a sensor module, and processing unit 110 may be configured to use and / or adapt the statistical model based in the real-time sensor data.

[0098] This real-time adjustment of the statistical model and therefore of the determined trajectory may narrow the gap between the model's initial predictions of event- related metrics and the actual metrics observed during the event. This adaptive processensures that the trajectory dynamically adjusts to changing conditions and unforeseen events as they occur, ultimately enhancing accuracy. As a result of these real-time adjustments, the accuracy of the predicted trajectory may vastly surpass a predetermined threshold, reaching levels of precision approaching 100%.

[0099] Once the new event is completed, all the tracking data (e.g., sensor data, movement analysis data) collected may serve as historical event data to update or generate a statistical model via the initialization process 200a, may be uploaded to a server, or may be used for performance analysis.

[0100] At step 216 in FIG. 2B, processing unit 110 may receive, via an external controller, one or more commands to modify or adjust the automatic tracking (i.e., the sequence of image data capture). For example, referring to FIGS. 1 A and 1 B, processing unit 110 may receive one or more commands from external controller 170 to modify or adjust the sequence of image data capture determined by tracking scheduler 160 (predictive tracking process). In some embodiments, external controller 170 may include a device or system external to processing unit 110 that is configured to issue commands, instructions, or signals (e.g., feedback signals) to control and manipulate the operation (e.g., predictive tracking operation) of processing unit 110 or system 100. For example, external controller 170 may be configured to send commands for adjusting camera parameters, including zoom level, focus, and frame rate, to enhance object tracking precision.

[0101] In certain embodiments, commands received by external controller 170 may include instructions to either speed up or slow down the sequence set by tracking scheduler 160. For example, these commands might dictate increasing or decreasing the determined sequence by a specified percentage (e.g., 2%, 5%, etc.) if the predictive tracking lags behind the event's progression. Such refinements or speed controladjustments may also be helpful when unforeseen events lead to the predicted trajectory's accuracy dropping below a predetermined threshold.

[0102] External controller 170 may be implemented with any suitable combination of hardware, software, and / or firmware, and included components such as a remote control, a computer interface, a mobile device, and / or other mechanisms capable of communicating with processing unit 110. The operations of external controller 170 may be automated through the use of a processor and executable instructions for sending commands to modify or adjust the tracking process at a venue. In some embodiments, external controller 170 may be further configured to receive input from a user via a control interface allowing manual modification of the tracked objects' trajectories in real-time. For example, referring to FIGS. 1 A and 1 B, external controller 170 may receive input from an operator or other user via control interface 175. Control interface 175 encompasses any system capable of providing a control / feedback signal, such as a user interface (III), a physical button, lever, touchscreen, voice command interface, motion sensor, or remote control device. In some embodiments, external controller 170 may serve as an interface through which operators, other users, or other systems interact with and exert control over processing unit 110 and / or system 100. It allows for remote operation and the adjustment of settings, initiation of actions, and / or application of changes to the tracking system's behavior without the need to directly interact with processing unit 110 or system 100 itself. In some embodiments, external controller 170 may include a computer-vision system. For example, referring to FIG. 1 B, computer-vision system 165 may be included in external controller 170 and may provide feedback signals to tracking scheduler 160 to modify the determined trajectory or adjust the speed of the predicted progression of the event (e.g., increasing or decreasing the determined sequence by a specified percentage). Additionally, or alternatively, external controller 170 may include a sensor module configured to collect and analyze real-time sensor data.

[0103] In some embodiments, the one or more commands received from external controller 170 may include adjustments to the statistical model parameters. This functionality allows operators or users to manually fine-tune the statistical model based on their preferences or specific requirements, either before the commencement of an event or in real-time during an ongoing tracking process for an event. For instance, operators or users may manually modify the initial conditions of the statistical model to better align with the current event dynamics or adjust various parameters on the fly to enhance the accuracy of object tracking. These adjustments empower operators or users to tailor the statistical model to suit specific scenarios, account for environmental factors, or adapt to changing conditions seamlessly. Whether optimizing initial conditions for better trajectory predictions or dynamically adjusting parameters to improve tracking precision, this flexibility enables operators or users to exert greater control over the statistical model's performance and responsiveness in diverse tracking scenarios. For example, a user via external controller 170 may specify prior to the commencement of an event, a level of the event (e.g., final race, qualifying race), weather conditions, or any other relevant parameters that may improve the predictive outcome of the statistical model.

[0104] In some embodiments, the external controller may be localized at the venue or provided remotely from the venue. For example, FIG. 8A is a diagrammatic representation of a first exemplary system configuration wherein external controller 170 is depicted as situated within venue 300. This setup allows operators present at the venue, to transmit input to system 100 through external controller 170 in response to scenes unfolding in real-time.. Operators and potentially other users who have a firsthand view of the proceedings, may provide real-time feedback or corrections to system 100 through external controller 170. By leveraging their direct observation and understanding of the situation, operators may contribute to refining the accuracy and fidelity of thesystem's tracking capabilities. Further, the presence of an external controller at the venue can enhance the system's adaptability and responsiveness to dynamic conditions, ensuring that it can effectively capture and interpret real-world events as they unfold.

[0105] By way of a further example, in FIG. 8B, a diagrammatic representation is provided of a second exemplary system configuration in which external controller 170 is positioned remotely, outside of any specific venue. In some embodiments, external controller 170 may be hosted on a server or cloud system and communicate with one or more venues via a communication network (e.g., the Internet or a Wide Area Network (WAN) or Local Area Network (LAN) system). In this setup, external controller 170 possesses the capability to communicate with multiple systems, denoted as system 100- a through 100-c, located at distinct venues labeled as venue 100-a through 100-c. This configuration can provide several notable advantages. Firstly, it eliminates the necessity for a distinct operator to be present at each venue. Instead, a single operator can effectively manage and interact with multiple tracking systems across different venues through the remote external controller. Further, external controller 170 can function as a centralized control mechanism that streamlines the management process, reducing the need for redundant personnel and resources at each individual venue. Moreover, this setup mitigates logistical challenges associated with physical presence at multiple venues. Attending events co-occurring or in close temporal proximity across geographically distant venues may be cumbersome and impractical. By centralizing control through a remote external controller, the need for operators to physically travel between venues is eliminated, overcoming barriers posed by time constraints and geographic distances for event management. The ability to communicate with multiple systems across distinct venues also enhances coordination, collaboration, and efficiency in managing and overseeing events or activities occurring simultaneously across different locations. Additionally, with a centralized external controller utilized for different events, itmay potentially be used for diverse event types. This flexibility underscores the versatility of the remote control setup, allowing a single external controller to interface with systems across various venues hosting a range of events.

[0106] Referring again to FIG. 2B, at step 218, a broadcast feed may be generated comprising the captured image data. Such a broadcast feed may be generated by processing unit 110 by sending control signal to a switcher (e.g., switcher 150 in FIG. 1A) or by processing and compiling captured image data from the set of cameras. For example, referring to FIG. 1 B, processing unit 110 may generate broadcast feed 180 based on the captured image data received from the set of cameras 155. In some embodiments, the broadcast feed may be transmitted in real-time to one or more remote devices for remote monitoring, analysis, or audience engagement. For example, broadcast feed 180 may be transmitted in real-time to remote devices 190. Remote devices 190 may encompass any sort of system capable of receiving a broadcast feed. Examples of remote devices include but are not limited to, one or more displays situated at the venue, one or more displays receiving the broadcast feed via a TV network and / or via Internet Protocols for live streaming, mobile devices, computers, or VR headsets. Additionally, or alternatively, in some embodiments, wherein the broadcast feed may be stored. For example, broadcast feed 180 may be uploaded to a server, stored in a memory device associated with processing unit 110, or stored in a permanent or nonpermanent storage communicating with system 100. This storage of broadcast feed 180 enables delayed diffusion, replay and review, on-demand access redundancy, and backup or content distribution across one or more remote devices such as remote devices 190.

[0107] In certain embodiments, the broadcast feed may be assembled using captured image data from a selected subset of the available cameras within the venue. For instance, while certain cameras are strategically positioned for tracking purposes,they may not offer compelling or relevant perspectives for spectators. Consequently, their feeds may be excluded from the broadcast feed. Alternatively, the feeds from multiple cameras may be redundant, and the broadcast feed might only require featuring one view at any given time. In some embodiments, the generation of the broadcast feed may follow a predetermined sequence, ensuring a smooth transition between different camera views. Moreover, in some cases, processing unit 110 may be further configured to select or sequence between multiple camera feeds to provide a cohesive and seamless viewing experience. In certain embodiments, processing unit 110 may be further configured to receive one or more commands via external controller 170, allowing for modifications or edits to the broadcast feed in real-time, thereby accommodating specific preferences or requirements of the viewing audience.

[0108] In some embodiments, the broadcast feed may be augmented with augmented reality (AR) elements, providing additional information or interactive features to users viewing the event through AR-enabled devices. This enhances the overall user experience and opens up new possibilities for engaging audiences during events as AR overlays can provide viewers with immersive and contextually relevant information as they observe the event. For instance, the broadcast feed may incorporate data derived from the tracking of one or more objects. As the event unfolds, real-time metrics such as instantaneous speed, acceleration, or positions in the race of the objects or participants can be dynamically displayed in proximity to the corresponding objects or participants. Furthermore, in some embodiments, the broadcast feed may comprise metadata indicating the confidence levels or uncertainties associated with the tracked object positions. This metadata may serve to augment the broadcast feed with valuable information regarding the reliability and accuracy of the tracking data. This inclusion of metadata may offer several benefits such as enhanced interpretations, decision-making support (e.g., for coaches or participant staff), and or quality assessment of the tracking.

[0109] In some embodiments, the automatic tracking may be synchronized with audio data capturing ambient sounds within the venue, providing a multi-modal representation of the event. For instance, any of the cameras depicted in FIG. 6 may be paired with a microphone designed to capture audio data simultaneously with the image data. In some embodiments, this synchronized audio data stream may then be seamlessly integrated into the broadcast feed to enrich the content and provide viewers with a more immersive experience. Additionally, in some embodiments, the multi-modal representation of the event may be further reinforced by integrating data received from one or more sensors attached to the objects / participants engaged in the event or localized at the venue. Integrating sensor data into the multi-modal representation may enhance the accuracy, granularity, and richness of the event feed. Viewers and analysts gain access to a wealth of real-time information that enhances their understanding and engagement with the event. Whether tracking the performance of athletes, assessing environmental conditions, or monitoring the behavior of objects, the fusion of sensor data with other media content may create a holistic view of the event that captures its complexity and nuance.

[0110] In some embodiments, processing unit 110 via computer-vision system 165 may be further configured to generate predictive alerts based on deviations between the determined trajectory and the actual movement of the tracked objects. These deviations may occur due to unexpected circumstances during an event, such as a collision between participants, a runner or racer stumbling, a car veering off the track or experiencing a malfunction, or a skier missing a gate. In such instances, the statistical model may fail to accurately predict the impact of such deviations and their subsequent outcomes. To address these uncertainties, predictive alerts may be triggered to alert an operator and / or controller system of potential deviations from the determined trajectories. Predictive alerts may serve as proactive indicators of potential disruptions or anomalies in the eventproceedings, enabling timely intervention or adjustment of tracking parameters. These alerts may take various forms, such as visual notifications, audible alarms, or system messages, depending on the operator's preferences and the seventy of the deviation. Upon receiving a predictive alert, the operator or controller system may take appropriate action to mitigate the impact of the deviation. For instance, the operator may provide manual commands via the external controller to adjust the tracking parameters, recalibrate the statistical model, and / or temporarily pause the tracking process to assess the situation. This proactive approach empowers operators to maintain control over the tracking process and adapt dynamically to changing event conditions. If deviations in trajectories pose a safety threat to the objects / participants of the event, predictive alerts may be further used to notify competent authorities (e.g., event organizers, safety personnel, race officials, or emergency responders) of potential hazards.

[0111] In addition to notifying competent authorities about safety threats, the alerts may also indicate deviations that represent forbidden acts based on event regulations and / or safety considerations. This ensures that the tracking system aligns with the rules and guidelines set by event authorities for fair play, safety, and event integrity. When deviations from expected trajectories suggest a violation of event regulations (e.g., false start, forbidden overpass, deviations from the regulated path or track, etc.) or safety protocols, the system generates alerts to highlight these infractions and prompt appropriate action. Moreover, in the context of animal racing, the tracking strategy can be adjusted based on safety protocols for the animals (horses, greyhounds, etc.). This ensures humane and ethical tracking practices, aligning with guidelines and regulations set for the protection of animals during events on racetracks or other venues. Such implementation of regulations on the tracking system may be performed via the external controller, where operators can input specific parameters or commands to ensure compliance with animal welfare standards and event regulations. By integrating safetyprotocols for animals into the tracking system, organizers can uphold ethical standards and prioritize the well-being of animals participating in events.

[0112] In some embodiments, the set of cameras may be configured not only to capture image data during the new event for tracking purposes but also to capture image data before and after the event without any tracking or along different determined trajectories. The generated broadcast feed may further include this additional image data, offering a comprehensive view of the entire event timeline. Examples of events occurring outside the main event and not requiring tracking, or tracking along a different trajectory include, but are not limited to, parade shots, podium shots, warm-up shots, national anthem shots, team hurdle shots, and coach shots. The additional shots or image data may provide context, atmosphere, and behind-the-scenes insights into the event, enhancing viewer engagement and understanding. Additionally, or alternatively, certain situations may occur during the event but are not directly related to the main proceedings. The set of cameras may also capture image data of such situations for inclusion in the broadcast feed. For instance, shots or image data of a car performing a pit stop may offer valuable glimpses into the intricacies and dynamics of the event. By incorporating such image data, the broadcast feed becomes more comprehensive, capturing not only the main event but also the surrounding activities and moments that contribute to the overall event experience. This approach may enrich the viewing experience by providing viewers with a holistic perspective of the event, encompassing both the main highlights and the peripheral activities. It offers a deeper immersion into the event atmosphere and fosters a greater appreciation for the intricacies and dynamics at play. By leveraging the capabilities of the set of camera systems to capture a wide range of shots and scenarios, the broadcast feed becomes more engaging, informative, and compelling for viewers.

[0113] The embodiments disclosed herein are exemplary and do not limit implementations of the claimed invention or any of the features thereof. As will beappreciated from this disclosure, modifications and adaptations may be made to the disclosed embodiments for tracking one or more objects during an event within a venue or other environments.

[0114] The foregoing description has been presented for purposes of illustration. It is not exhaustive and not limited to the precise forms or embodiments disclosed. Modifications and adaptations will be apparent to those skilled in the art from consideration of the specification and practice of the disclosed embodiments. Additionally, although aspects of the disclosed embodiments are described as being stored in memory, one skilled in the art will appreciate that these aspects can also be stored on other types of computer readable media, such as secondary storage devices, for example, hard disks or CD ROM, or other forms of RAM or ROM, USB media, DVD, Blu-ray, 4K Ultra HD Blu-ray, or other optical drive media.

[0115] Computer programs based on the written description and disclosed methods are within the skill of an experienced developer. The various programs or program modules can be created using techniques known to one skilled in the art or can be designed in connection with existing software. For example, program sections or program modules can be designed in or by means of .Net Framework, .Net Compact Framework (and related languages, such as Visual Basic, C, etc.), Java, C++, Objective- C, HTML, HTML / AJAX combinations, XML, or HTML with included Java applets.

[0116] Moreover, while illustrative embodiments have been described herein, the scope of any and all embodiments having equivalent elements, modifications, omissions, combinations (e.g., of aspects across various embodiments), adaptations and / or alterations as would be appreciated by those skilled in the art based on the present disclosure. The limitations in the claims are to be interpreted broadly based on the language employed in the claims and not limited to examples described in the present specification or during the prosecution of the application. The examples are to beconstrued as non-exclusive. Furthermore, the steps of the disclosed methods may be modified in any manner, including by reordering steps and / or inserting or deleting steps. It is intended, therefore, that the specification and examples be considered as illustrative only, with a true scope and spirit being indicated by the following claims and their full scope of equivalents.

Claims

CLAIMS:1 . A computer-implemented method for tracking objects engaged in an event within a venue, the method comprising steps performed by at least one processor configured with executable instructions, the steps including: providing a statistical model based on historical event data for at least one event type; providing a 3-dimensional (3D) model of a venue using map data; determining a trajectory within the venue based on the 3D and statistical models; determining a sequence of image data capture for a set of cameras installed at the venue based on the determined trajectory, wherein the sequence enables one or more objects engaged in an event to be automatically tracked along the determined trajectory via the set of cameras installed at the venue, and wherein image data of the one or more objects are captured by the set of cameras in the determined sequence; receiving, via an external controller, one or more commands to modify the automatic tracking; and generating a broadcast feed comprising the captured image data.

2. The computer-implemented method of claim 1 , wherein the historical event data for the at least one event type includes information related to object movements and interactions during past occurrences of the event type.

3. The computer-implemented method of claim 1 , wherein the 3D model of the venue comprises one or more event paths.

4. The computer-implemented method of claim 1 , further comprising adapting the statistical model in real-time based on live data obtained during the event.

5. The computer-implemented method of claim 1 , wherein determining the trajectory comprises determining the trajectory of the one or more objects by dynamically adjusting the statistical model based on real-time sensor data.

6. The computer-implemented method of claim 1 , wherein the set of cameras includes a combination of fixed and movable cameras, and the method further comprises controlling and adjusting at least one of an orientation and a position of the movable cameras according to the determined trajectory.

7. The computer-implemented method of claim 6, wherein the movable cameras include preconfigured camera paths, and the method further comprises adjusting the preconfigured camera paths according to the determined trajectory.

8. The computer-implemented method of claim 1 , further comprising generating predictive alerts based on deviations between the determined trajectory and the actual movement of the tracked objects.

9. The computer-implemented method of claim 1 , wherein the external controller is configured to send commands for adjusting camera parameters, including zoom level, focus, and frame rate, to enhance object tracking precision.

10. The computer-implemented method of claim 1 , wherein the external controller is configured to send commands for either speed up or slow down the determined sequence of image data capture.

11. The computer-implemented method of claim 1 , wherein the external controller is further configured to receive input from a user interface allowing manual modification of the trajectory of the tracked objects in real-time.

12. The computer-implemented method of claim 1 , further comprising combining the 3D model and the statistical model.

13. The computer-implemented method of claim 12, wherein a location of each camera of the set of cameras is mapped into the combined 3D and statistical models.

14. The computer-implemented method of claim 1 , wherein determining the trajectory comprises determining the trajectory based on at least one of optimizing camera coverage, minimizing blind spots, and maximizing the quality of captured image data.

15. The computer-implemented method of claim 1 , wherein determining the trajectory comprises determining a path based on the 3D model and determining the trajectory based on the determined path and the statistical model.

16. The computer-implemented method of claim 1 , wherein determining the trajectory comprises determining a preliminary path based on the 3D model, refining the preliminary path based on the statistical model and determining the trajectory based on the refined preliminary path and the statistical model.

17. The computer-implemented method of claim 1 , wherein the set of cameras includes infrared or low-light cameras to facilitate tracking in low-visibility conditions.

18. The computer-implemented method of claim 1 , wherein the set of cameras includes PTZ cameras.

19. The computer-implemented method of claim 1 , further comprising synchronizing the automatic tracking with audio data capturing of ambient sounds within the venue, and providing a multi-modal representation of the event.

20. The computer-implemented method of claim 1 , wherein the broadcast feed is transmitted in real-time to one or more remote devices for at least one of monitoring, analysis, or audience engagement.

21. The computer-implemented method of claim 1 , wherein the broadcast feed is stored in a memory.

22. The computer-implemented method of claim 1 , wherein the external controller is localized at the venue or remotely from the venue.

23. The computer-implemented method of claim 1 , wherein the set of cameras is further configured to capture image data before and after the event without any tracking or tracking along different determined trajectories, and the generated broadcast feed further includes the image data captured before and after the event.

24. The computer-implemented method of claim 1 , further comprising providing one or more initial conditions prior to the commencement of the event.

25. A non-transitory computer-readable medium containing instructions that when executed by at least one processor cause the at least one processor to perform operations for tracking objects engaged in an event within a venue, the operations comprising: providing a statistical model based on historical event data for at least one event type;providing a 3-dimensional (3D) model of a venue using map data; determining a trajectory within the venue based on the combined 3D and statistical models; determining a sequence of image data capture for a set of cameras installed at the venue based on the determined trajectory, wherein the sequence enables one or more objects engaged in an event to be automatically tracked along the determined trajectory via the set of cameras installed at the venue, and wherein image data of the one or more objects are captured by the set of cameras in the determined sequence; receiving, via an external controller, one or more commands to modify the automatic tracking; and generating a broadcast feed comprising the captured image data.

26. The non-transitory computer-readable medium of claim 25, wherein the historical event data for the at least one event type includes information related to object movements and interactions during past occurrences of the event type.

27. The non-transitory computer-readable medium of claim 25, wherein the 3D model of the venue comprises one or more event paths.

28. The non-transitory computer-readable medium of claim 25, wherein the operations further comprise adapting the statistical model in real-time based on live data obtained during the event.

29. The non-transitory computer-readable medium of claim 25, wherein determining the trajectory comprises determining the trajectory of the one or more objects by dynamically adjusting the statistical model based on real-time sensor data.

30. The non-transitory computer-readable medium of claim 25, wherein the set of cameras includes a combination of fixed and movable cameras, and the method further comprises controlling and adjusting at least one of an orientation and a position of the movable cameras according to the determined trajectory.

31. The non-transitory computer-readable medium of claim 30, wherein the movable cameras include preconfigured camera paths, and the method further comprises adjusting the preconfigured camera paths according to the determined trajectory.

32. The non-transitory computer-readable medium of claim 25, wherein the operations further comprise generating predictive alerts based on deviations between the determined trajectory and the actual movement of the tracked objects.

33. The non-transitory computer-readable medium of claim 25, wherein the external controller is configured to send commands for adjusting camera parameters, including zoom level, focus, and frame rate, to enhance object tracking precision.

34. The non-transitory computer-readable medium of claim 25, wherein the external controller is configured to send commands for either speed up or slow down the determined sequence of image data capture.

35. The non-transitory computer-readable medium of claim 25, wherein the external controller is further configured to receive input from a control interface allowing manual modification of the trajectory of the tracked objects in real-time.

36. The non-transitory computer-readable medium of claim 25, wherein the operations further comprise combining the 3D model and statistical model.

37. The non-transitory computer-readable medium of claim 36, wherein a location of each camera of the set of cameras is mapped into the combined 3D and statistical models.

38. The non-transitory computer-readable medium of claim 25, wherein determining the trajectory comprises determining the trajectory based on at least one of optimizing camera coverage, minimizing blind spots, and maximizing the quality of captured image data.

39. The non-transitory computer-readable medium of claim 25, wherein determining the trajectory comprises determining a path based on the 3D model and determining the trajectory based on the determined path and the statistical model.

40. The non-transitory computer-readable medium of claim 25, wherein determining the trajectory comprises determining a preliminary path based on the 3D model, refining the preliminary path based on the statistical model and determining the trajectory based on the refined preliminary path and the statistical model.

41. The non-transitory computer-readable medium of claim 25, wherein the set of cameras includes infrared or low-light cameras to facilitate tracking in low-visibility conditions.

42. The non-transitory computer-readable medium of claim 25, wherein the set of cameras includes PTZ cameras.

43. The non-transitory computer-readable medium of claim 25, wherein the operations further comprise synchronizing the automatic tracking with audio data capturing of ambient sounds within the venue, and providing a multi-modal representation of the event.

44. The non-transitory computer-readable medium of claim 25, wherein the broadcast feed is transmitted in real-time to one or more remote devices for at least one of monitoring, analysis, or audience engagement.

45. The non-transitory computer-readable medium of claim 25, wherein the broadcast feed is stored in a memory.

46. The non-transitory computer-readable medium of claim 25, wherein the external controller is localized at the venue or remotely from the venue.

47. The non-transitory computer-readable medium of claim 25, wherein the set of cameras is further configured to capture image data before and after the event without any tracking or tracking along different determined trajectories, and the generated broadcast feed further includes the image data captured before and after the event.

48. The non-transitory computer-readable medium of claim 25, wherein the operations further comprise providing one or more initial conditions prior to the commencement of the event.

49. A system for tracking objects engaged in an event within a venue, the system comprising: at least one processor configured with executable instructions to: provide a statistical model based on historical event data for at least one event type; provide a 3-dimensional (3D) model of a venue using map data; determining a trajectory within the venue based on the combined 3D and statistical models;determining a sequence of image data capture for a set of cameras installed at the venue based on the determined trajectory, wherein the sequence enables one or more objects engaged in an event to be automatically tracked along the determined trajectory via the set of cameras installed at the venue, and wherein image data of the one or more objects are captured by the set of cameras in the determined sequence; receive, via an external controller, one or more commands to modify the automatic tracking; and generate a broadcast feed comprising the captured image data.

50. The system of claim 49, wherein the historical event data for the at least one event type includes information related to object movements and interactions during past occurrences of the event type.51 . The system of claim 49, wherein the 3D model of the venue comprises one or more event paths.

52. The system of claim 49, wherein the at least one processor is further configured to adapt the statistical model in real-time based on live data obtained during the event.

53. The system of claim 49, wherein determining the trajectory comprises determining the trajectory of the one or more objects by dynamically adjusting the statistical model based on real-time sensor data.

54. The system of claim 49, wherein the set of cameras includes a combination of fixed and movable cameras, and the method further comprises controlling and adjusting at least one of an orientation and a position of the movable cameras according to the determined trajectory.

55. The system of claim 54, wherein the movable cameras include preconfigured camera paths, and the method further comprises adjusting the preconfigured camera paths according to the determined trajectory.

56. The system of claim 49, wherein the at least one processor is further configured to generate predictive alerts based on deviations between the determined trajectory and the actual movement of the tracked objects.

57. The system of claim 49, wherein the external controller is configured to send commands for adjusting camera parameters, including zoom level, focus, and frame rate, to enhance object tracking precision.

58. The system of claim 49, wherein the external controller is is configured to send commands for either speed up or slow down the determined sequence of image data capture.

59. The system of claim 49, wherein the external controller is further configured to receive input from a control interface allowing manual modification of the trajectory of the tracked objects in real-time.

60. The system of claim 49, wherein the at least one processor is further configured to combine the 3D model and the statistical model.61 . The system of claim 60, wherein a location of each camera of the set of cameras is mapped into the combined 3D and statistical models.

62. The system of claim 49, wherein determining the trajectory comprises determining the trajectory based on at least one of optimizing camera coverage, minimizing blind spots, and maximizing the quality of captured image data.

63. The system of claim 49, wherein determining the trajectory comprises determining a preliminary path based on the 3D model, refining the preliminary path based on the statistical model and determining the trajectory based on the refined preliminary path and the statistical model.

64. The system of claim 49, wherein the at least one processor is further configured to train the statistical model using a combination of supervised and unsupervised learning techniques to adapt to various event types and venue configurations.

65. The system of claim 49, wherein the set of cameras includes infrared or low-light cameras to facilitate tracking in low-visibility conditions.

66. The system of claim 49, wherein the set of cameras includes PTZ cameras.

67. The system of claim 49, wherein the at least one processor is further configured to synchronize the automatic tracking with audio data capturing of ambient sounds within the venue, and providing a multi-modal representation of the event.

68. The system of claim 49, wherein the broadcast feed is transmitted in real-time to one or more remote devices for at least one of monitoring, analysis, or audience engagement.

69. The system of claim 49, wherein the broadcast feed is stored in a memory.

70. The system of claim 49, wherein the external controller is localized at the venue or remotely from the venue.71 . The system of claim 49, wherein the set of cameras is further configured to capture image data before and after the event without any tracking or tracking along different determined trajectories, and the generated broadcast feed further includes the image data captured before and after the event.

Images