Real target flying disc gun simulation shooting system and method based on light sensing tracking and medium

Abstract

The application provides a real target flying disc gun simulation shooting system and method based on light sensing tracking, and a medium. The system comprises: a target throwing machine, a light sensing gun provided with a camera, a trigger unit and a processing unit, and the shooting direction of the camera is consistent with the orientation of the muzzle; a central control scheduling module for sending a target throwing signal to the target throwing machine and the light sensing gun; a control analysis processing module provided with a deep learning network model; the control analysis processing module determines the moving track data of the light sensing gun based on the real-time video stream obtained by the camera, obtains a visual scene image from the real-time video stream at the shooting moment and inputs the visual scene image into the deep learning network model, and determines a hit result based on the moving track data and the target space position of the flying disc. According to the technical scheme of the embodiment of the application, the deep learning network model can be used to perform image processing on the video stream shot by the camera, and a pure visual perception scheme is adopted to realize real-time detection, tracking and hit determination of the flying disc target.

Description

Technical Field

[0001] This invention relates to the field of simulated shooting technology, and in particular to a live-fire target skeet gun simulated shooting system, method and medium based on light-sensing tracking. Background Technology

[0002] As an official competition event, skeet shooting traditionally uses live-fire shooting in both training and competition. Live-fire shooting has the following drawbacks: First, it poses safety hazards, requiring dedicated shooting ranges and strict safety management measures; second, training costs are high, with large consumption of bullets and skeet targets; third, training effectiveness is difficult to quantify and evaluate, making it impossible to accurately analyze and scientifically guide shooters' techniques.

[0003] To overcome the shortcomings of live-fire shooting, simulated shooting technology has emerged. Current technology primarily employs laser simulation, where a laser emitter is mounted on the gun. When the trigger is pulled, a laser is emitted, and a photoelectric sensor captures the position of the laser spot for hit determination. However, this approach suffers from the following technical bottlenecks: 1. System delays cause time asynchrony; there is a time difference between laser emission and acquisition, and since the flying saucer is in high-speed motion (typically 20-30 m / s), millisecond-level delays can lead to centimeter-level positional errors; 2. Mechanical vibrations during firing cause laser spot trailing, severely affecting coordinate positioning accuracy; 3. It is impossible to completely record the muzzle dynamic trajectory, making quantitative analysis of aiming techniques difficult. Summary of the Invention

[0004] This invention aims to at least solve one of the technical problems existing in the prior art. To this end, this invention proposes a live-target UFO gun simulation shooting system, method, and medium based on optical tracking, which can achieve real-time detection, tracking, and hit determination of UFO targets using a purely visual perception scheme.

[0005] In a first aspect, embodiments of the present invention provide a live-target skeet gun simulation shooting system based on optical tracking, comprising: Multiple target launchers, responding to target launch signals, launch flying saucers; A light-sensitive gun has a muzzle and a trigger. A camera, a trigger unit, and a processing unit are arranged on the upper side of the gun body. The camera and the processing unit are electrically connected, and the trigger unit is electrically connected to the trigger and the processing unit. The shooting direction of the camera is the same as the direction of the muzzle. The central control and dispatch module is used to send the target launching signal to at least one of the target launching machines and the light-sensing gun; A control analysis and processing module is communicatively connected to the processing unit, and the control analysis and processing module is pre-configured with a deep learning network model. The processing unit responds to the target throwing signal to start the camera to acquire a real-time video stream and send it to the control analysis and processing module, and responds to the trigger signal generated by the trigger unit to control the camera to stop shooting; The control analysis and processing module determines the movement trajectory data of the light-sensing gun based on the real-time video stream, determines the firing time based on the trigger signal, acquires a visual scene image from the real-time video stream based on the firing time and inputs it into the deep learning network model, determines the target spatial position of the flying saucer based on the visual scene image through the deep learning network model, and determines the hit result based on the movement trajectory data and the target spatial position.

[0006] According to some embodiments of the present invention, the gun body of the light-sensing gun is further provided with a loading unit, a power supply unit, an audio unit and a communication unit, the processing unit is electrically connected to the loading unit, the power supply unit, the audio unit and the communication unit respectively, and the communication unit is communicatively connected to the control analysis and processing module.

[0007] According to some embodiments of the present invention, it further includes: The tablet performance recording module is communicatively connected to the control analysis and processing module. The tablet performance recording module is used to display the shooting scores and training data analysis results of each group in real time. The shooting scores and training data analysis results are generated by the control analysis and processing module. The network communication module includes a router and a Bluetooth device. The central control and scheduling module, the tablet score recording module, and the light sensor are connected to each other through the network communication module.

[0008] Secondly, embodiments of the present invention also provide a method for simulating live-fire clay pigeon shooting based on light-sensing tracking, applied to the live-fire clay pigeon shooting simulation system based on light-sensing tracking as described in the first aspect, the method comprising: The central control and dispatch module sends a target launching signal to at least one target launching machine and a light sensor gun, wherein the target launching machine launches the flying saucer in response to the target launching signal, and the light sensor gun activates the camera to acquire a real-time video stream in response to the target launching signal; The control analysis and processing module determines the movement trajectory data of the light sensor gun based on the real-time video stream, determines the firing time based on the trigger signal of the light sensor gun, and acquires a visual scene image from the real-time video stream based on the firing time and inputs it into the deep learning network model. The deep learning network model determines the target spatial location of the UFO based on the visual scene image, and determines the hit result based on the movement trajectory data and the target spatial location.

[0009] According to some embodiments of the present invention, the deep learning network model includes a backbone network, a neck network, and a detection head. Determining the target spatial location of the UFO based on the visual scene image using the deep learning network model includes: The visual scene image is input into the backbone network, and multiple semantic feature maps at different levels are extracted through the backbone network. The semantic feature map is upsampled and fused using the neck network to obtain multiple enhanced feature maps at different levels; Each of the enhanced feature maps is input into the detection head to determine its corresponding preset anchor point position and corresponding confidence score, and the target spatial position is determined based on the preset anchor point position.

[0010] According to some embodiments of the present invention, determining the target spatial location based on the preset anchor point location includes: Determine the geometric parameters of each preset anchor point position relative to the target bounding box, wherein the geometric parameters include the offset of the target bounding box relative to the preset anchor point, the width scaling, the height scaling, and the rotation angle; Based on the geometric parameters, the initial rotation rectangle of the flying saucer in the preset model coordinate system is decoded; Based on the corresponding confidence scores and preset confidence thresholds, multiple candidate rotated rectangles are determined from the multiple preliminary rotated rectangles; The target rotating rectangle is determined from multiple candidate rotating rectangles based on a nonmaximum suppression algorithm; The coordinate information of the target rotating rectangle in the model coordinate system is used to determine the target spatial position.

[0011] According to some embodiments of the present invention, determining the hit result based on the movement trajectory data and the target spatial location includes: The target hit range is determined based on the currently set target difficulty. The control analysis and processing module has multiple preset difficulty levels, and the preset difficulty level is inversely proportional to the corresponding preset hit range. The hit result is determined based on the target hit range and the target spatial location.

[0012] According to some embodiments of the present invention, the light-tracking-based live-target clay pigeon shooting simulation system further includes a tablet-based score recording module. After determining the hit result based on the movement trajectory data and the target's spatial position, the method further includes: After receiving the target launch signal, a timer is started, and the moment when the flying saucer enters the target's hit range is determined as the lock-on moment; The following time is determined based on the time interval between the locking moment and the firing moment; A shooting score is generated based on the hit result, and the training data analysis result is determined based on the follow-up time and the hit result; The shooting scores and training data analysis results are sent to the tablet score recording module for real-time display via the control analysis and processing module.

[0013] Thirdly, embodiments of the present invention provide a live-fire target skeet gun simulation shooting system based on light-sensing tracking, including at least one control processor and a memory for communicatively connecting with the at least one control processor; the memory stores instructions executable by the at least one control processor, which are executed by the at least one control processor to enable the at least one control processor to perform the live-fire target skeet gun simulation shooting method based on light-sensing tracking as described in the second aspect above.

[0014] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing computer-executable instructions for executing the light-tracking-based simulated firing method for a live-target skeet gun as described in the second aspect above.

[0015] The optical tracking-based live-target skeet gun simulation shooting system according to embodiments of the present invention has at least the following beneficial effects: multiple target launchers launch skeets in response to a target launch signal; an optical gun has a muzzle and a trigger on its body, and a camera, a trigger unit, and a processing unit are disposed on the upper side of the gun body, the camera and the processing unit are electrically connected, the trigger unit is electrically connected to the trigger and the processing unit, and the shooting direction of the camera is consistent with the direction of the muzzle; a central control scheduling module is used to send the target launch signal to at least one of the target launchers and the optical gun; a control analysis and processing module is communicatively connected to the processing unit, and the control analysis and processing module is pre-loaded with a deep learning network. The deep learning network model is described in the following manner: The processing unit, in response to a target-launching signal, activates the camera to acquire a real-time video stream and sends it to the control and analysis processing module; the trigger unit, in response to a trigger signal generated by the trigger unit, controls the camera to stop recording. The control and analysis processing module determines the movement trajectory data of the light-sensing gun based on the real-time video stream, determines the firing time based on the trigger signal, acquires a visual scene image from the real-time video stream based on the firing time and inputs it into the deep learning network model. The deep learning network model then determines the target spatial position of the flying saucer based on the visual scene image, and determines the hit result based on the movement trajectory data and the target spatial position. According to the technical solution of this embodiment, a deep learning network model can be used to perform image processing on the video stream captured by the camera, and a pure visual perception scheme can be used to achieve real-time detection, tracking, and hit determination of the flying saucer target. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a live-target flying saucer gun simulation shooting system based on light-sensing tracking provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a light-sensing gun module provided in another embodiment of the present invention; Figure 3 This is a flowchart of a live-target flying saucer simulated shooting method based on light-sensing tracking provided in another embodiment of the present invention; Figure 4 This is a structural diagram of a live-target flying saucer gun simulation shooting system based on light-sensing tracking, provided in another embodiment of the present invention. Detailed Implementation

[0017] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0018] In the description of this invention, it should be understood that the orientation descriptions, such as up, down, front, back, left, right, etc., are based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention.

[0019] In the description of this invention, "several" means one or more, "more than" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0020] In the description of this invention, unless otherwise explicitly defined, terms such as "setting," "installing," and "connecting" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this invention in conjunction with the specific content of the technical solution.

[0021] This invention provides a live-target skeet gun simulation shooting system, method, and medium based on light-sensing tracking. The method includes: multiple target launchers that launch skeets in response to a launch signal; a light-sensing gun with a muzzle and trigger, a camera, a trigger unit, and a processing unit disposed on the upper side of the gun body, the camera and the processing unit being electrically connected, the trigger unit being electrically connected to the trigger and the processing unit, and the camera's shooting direction being consistent with the muzzle's orientation; a central control and scheduling module for sending the launch signal to at least one target launcher and the light-sensing gun; and a control and analysis processing module communicatively connected to the processing unit. The processing module is pre-configured with a deep learning network model. The processing unit, in response to a target-launching signal, activates the camera to acquire a real-time video stream and sends it to the control and analysis processing module. In response to a trigger signal generated by the trigger unit, it controls the camera to stop recording. The control and analysis processing module determines the trajectory data of the light-sensing gun based on the real-time video stream, determines the firing time based on the trigger signal, acquires a visual scene image from the real-time video stream based on the firing time, and inputs it into the deep learning network model. The deep learning network model determines the target spatial position of the flying saucer based on the visual scene image, and determines the hit result based on the trajectory data and the target spatial position. According to the technical solution of this embodiment, a deep learning network model can be used to process the video stream captured by the camera, achieving real-time detection, tracking, and hit determination of the flying saucer target using a pure visual perception scheme.

[0022] The technical solutions of the embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0023] Reference Figure 1 , Figure 1 This is a schematic diagram of a live-target clay pigeon shooting simulation system based on light-sensing tracking, provided by an embodiment of the present invention. The live-target clay pigeon shooting simulation system based on light-sensing tracking includes: Multiple target launchers, responding to target launch signals, launch flying saucers; The light-sensing gun has a muzzle and a trigger. A camera, a trigger unit, and a processing unit are located on the upper side of the gun body. The camera and the processing unit are electrically connected, and the trigger unit is electrically connected to the trigger and the processing unit. The shooting direction of the camera is the same as the direction of the muzzle. The central control and dispatch module is used to send target launching signals to at least one target launching machine and one light-sensing gun; The control analysis and processing module is connected to the processing unit and has a pre-set deep learning network model. The processing unit responds to the target throwing signal to start the camera to acquire real-time video stream and sends it to the control analysis and processing module, and responds to the trigger signal generated by the trigger unit to control the camera to stop shooting. The control analysis and processing module determines the movement trajectory data of the light sensor gun based on the real-time video stream, determines the firing time based on the trigger signal, acquires visual scene images from the real-time video stream based on the firing time and inputs them into the deep learning network model, determines the target spatial position of the UFO based on the visual scene images through the deep learning network model, and determines the hit result based on the movement trajectory data and the target spatial position.

[0024] It should be noted that there can be multiple target launchers, and the launch sequence, launch angle, and launch mode of multiple target launchers can be uniformly controlled by a central control system. The target launcher in this embodiment can be a civilian target launcher or a competition target launcher.

[0025] It should be noted that, as Figure 2 As shown, the light sensor gun has multiple unit modules. The trigger unit is connected to the trigger, and the trigger unit can generate a trigger signal after the trigger is pulled.

[0026] It should be noted that the central control and dispatch module can be a tablet or a laptop, without further limitation. After generating the target launching signal, the central control and dispatch module synchronously sends the target launching machine and the light sensor gun. The target launching machine launches the flying saucer in response to the target launching signal, and the light sensor gun activates the camera to take pictures in response to the target launching signal, providing a basis for subsequent determination of each moment.

[0027] It should be noted that the control analysis and processing module is pre-set with a deep learning network model, such as a common target detection model. Through the communication connection, the processing unit sends the real-time video stream captured by the camera to the control analysis and processing module in real time. Since the camera is oriented in the same direction as the muzzle, the camera's shooting trajectory can be equated to the trajectory of the light sensor gun. Since the environment of the shooting range is fixed, the trajectory data can be determined through simple image processing. The trajectory data is used to characterize the trajectory of the light sensor gun, providing a data foundation for subsequent determination of training analysis data and hit results.

[0028] It should be noted that after the trigger signal is obtained, the control analysis and processing module extracts the current original video frame as the visual scene image. It can determine the shooting position as the center position of the visual scene image, and determine the target spatial position of the UFO from the visual scene image through target detection. Based on the target spatial position, it determines whether the UFO has been hit. Thus, the hit result of the UFO is determined by a pure visual solution without consuming the physical UFO.

[0029] The control analysis and processing module is configured to perform the following steps: (1) Aiming process recording and uploading steps: In response to the target throwing signal, the camera of the light sensor gun is activated to start continuous recording, recording the entire aiming and firing process from the moment the flying saucer is thrown to the completion of the light sensor gun firing, generating a complete aiming process video file, and uploading the aiming process video file to the results display terminal; the aiming process video file is used to intuitively display and replay the movement trajectory and aiming dynamics of the light sensor gun throughout the shooting process on the results display terminal; (2) Visual playback step of aiming trajectory: The results display terminal is equipped with a video playback component, which supports adjustable speed playback of the video file of the aiming process, including slow motion, pause and frame-by-frame playback functions, so as to realize the fine visualization, tracing, analysis and review of the shooter's aiming action; (3) Visual perception and target localization steps: At the moment of firing, the target frame image corresponding to the moment of firing is extracted from the video file of the aiming process, and the target frame image is input into the deep learning UFO detection network model. The deep learning UFO detection network model outputs the pixel coordinates of the UFO in the target frame image as the target spatial position of the UFO. (4) Adjustable difficulty hit determination steps: Obtain the radius of the corresponding circular hit area according to the preset difficulty level. The difficulty level includes at least three levels: hard, normal, and easy. Each level corresponds to a unique and different circular hit area radius value, and the higher the difficulty level, the smaller the corresponding circular hit area radius value. Define a circular hit area in the image with the center point of the target frame image as the center and the radius of the circular hit area corresponding to the current difficulty level as the radius. When the pixel coordinates of the UFO are located inside the circular hit area, it is determined that the shot hits; otherwise, it is determined that it misses. (5) Hit result display step: Associate the hit judgment result with the corresponding aiming process video file, and synchronously display the hit result and the playback interface of the aiming process video file on the result display terminal.

[0030] Additionally, in one embodiment, such as Figure 2 As shown, the light sensor gun also has a loading unit, a power supply unit, an audio unit, and a communication unit. The processing unit is electrically connected to the loading unit, the power supply unit, the audio unit, and the communication unit, respectively. The communication unit is communicatively connected to the control analysis and processing module.

[0031] It should be noted that the loading unit is located on the loading lever of the gun body and is electrically connected to the processing unit. The loading signal generated by the loading unit simulates the loading operation of the light sensor gun.

[0032] It should be noted that the power supply unit is used to provide operating power to each unit, and will not be further limited here.

[0033] It should be noted that the audio unit can be used to play simulated shooting sounds, improving the user experience.

[0034] Additionally, in one embodiment, it further includes: The tablet-based score recording module is connected to the control analysis and processing module. The tablet-based score recording module is used to display the shooting scores and training data analysis results of each group in real time. The shooting scores and training data analysis results are generated by the control analysis and processing module. The networking communication module includes a router and a Bluetooth device. The central control and dispatch module, the tablet score recording module, and the light sensor are connected to each other through the networking communication module.

[0035] It should be noted that the tablet performance recording module can obtain shooting scores and training data analysis results. Shooting scores can be multiple hits obtained from multiple shots, and training data analysis results can be determined based on the movement trajectory data of each shot.

[0036] It should be noted that the networking communication module includes a router and a Bluetooth device, which are used to build intra-component communication for the light-tracking-based live-target skeet gun simulation shooting system, ensuring that the central control and scheduling module, the tablet score recording module, and the light-sensing gun are interconnected but do not interfere with each other.

[0037] Additionally, refer to Figure 3 The present invention also provides a method for simulating live-fire clay pigeon shooting based on light-sensing tracking, including but not limited to the following steps: S10, the central control and dispatch module sends a target launching signal to at least one target launching machine and a light sensor gun, wherein the target launching machine launches the flying saucer in response to the target launching signal, and the light sensor gun activates the camera to acquire a real-time video stream in response to the target launching signal; S20: The control analysis and processing module determines the movement trajectory data of the light sensor gun based on the real-time video stream, determines the firing time based on the trigger signal of the light sensor gun, and acquires the visual scene image from the real-time video stream based on the firing time and inputs it into the deep learning network model. S30 uses a deep learning network model to determine the target spatial location of the flying saucer based on visual scene images, and determines the hit result based on the movement trajectory data and the target spatial location.

[0038] It should be noted that the principle of the simulated shooting method can be referred to in the description of the simulated shooting system above, and will not be repeated here for the sake of simplicity.

[0039] In another embodiment, the deep learning network model includes a backbone network, a neck network, and a detection head. In step S30, the target spatial location of the UFO is determined based on the visual scene image using the deep learning network model, specifically including but not limited to the following steps: S311, input the visual scene image into the backbone network, and extract semantic feature maps at multiple different levels through the backbone network; S312 uses a neck network to upsample and fuse semantic feature maps to obtain multiple enhanced feature maps at different levels; S313, input each enhanced feature map into the detection head to determine the corresponding preset anchor point position and the corresponding confidence score, and determine the target spatial position based on the preset anchor point position.

[0040] It should be noted that when the trigger unit detects the action of the trigger, it generates a trigger signal, which enables the processing unit to directly obtain the original image frame at the firing moment from the real-time video stream. The original video frame is the visual scene image of this embodiment. The original image frame is the raw image data that is not compressed or converted in format, and its resolution is consistent with the camera configuration, such as 3840×2160 pixels, which serves as the direct input source for the deep learning neural network model.

[0041] It should be noted that in this embodiment, the captured single-frame raw image is input into an optimized lightweight object detection model for processing. This model adopts a single-stage detection architecture. After acquiring the raw image frame, it is first scaled to a fixed input size preset by the model, and pixel values ​​are normalized to form a standard three-dimensional tensor input. This tensor input first enters the backbone network of the model. The backbone network consists of multiple cascaded convolutional modules, with the core module being a CSP structure that incorporates cross-stage partial connectivity. Each convolutional module contains a convolutional layer, a batch normalization layer, and an activation function. The image data passes sequentially through these modules, undergoing multiple downsampling operations, with the spatial size gradually decreasing and the number of feature channels gradually increasing. This process gradually abstracts and extracts multi-level semantic feature maps from low-level edges and textures to high-level semantics (such as the shape and contour features of a flying saucer). These feature maps have different spatial resolutions, corresponding to receptive fields of different scales in the input image.

[0042] Then, multiple semantic feature maps output from the backbone network are fed into a feature pyramid network for fusion. This network bidirectionally fuses high-level semantic information from deep feature maps with high-resolution detail information from shallow feature maps through a top-down upsampling path and a bottom-up downsampling path. Specifically, small-sized deep feature maps are upsampled and concatenated with shallow feature maps of matching size; simultaneously, shallow features exchange information with deep features through downsampling. This process generates a new set of multi-scale enhanced feature maps that integrate rich contextual information and precise location information, significantly improving the model's ability to detect targets of different sizes, especially small-sized or distant UFO targets.

[0043] Then, each enhanced feature map after fusion is input in parallel into a decoupled detection head. In the classification branch of the detection head, the feature map is processed through a series of convolutional layers to predict the confidence score of the presence of a UFO target at each preset anchor point position on the feature map, thereby determining the target spatial location of the UFO.

[0044] In another embodiment, determining the target spatial location based on the preset anchor point location in step S313 further includes, but is not limited to, the following steps: S3131, determine the geometric parameters of each preset anchor point position relative to the target bounding box, wherein the geometric parameters include the offset of the target bounding box relative to the preset anchor point, the width scaling amount, the height scaling amount, and the rotation angle; S3132, decodes the initial rotation rectangle of the flying saucer in the preset model coordinate system based on geometric parameters; S3133, Based on the corresponding confidence scores and preset confidence thresholds, determine multiple candidate rotated rectangles from multiple preliminary rotated rectangles; S3134, the target rotating rectangle is determined from multiple candidate rotating rectangles based on the nonmaximum suppression algorithm; S3135, determine the target spatial position by the coordinate information of the target rotating rectangle in the model coordinate system.

[0045] It should be noted that the detection head in this embodiment also includes a regression branch. That is, the detection head includes independent classification and regression branches. The regression branch processes the same feature map through another series of convolutional layers to predict the geometric parameters of the target bounding box at each preset anchor point. Accurate fitting is achieved using a rotated bounding box. The regression parameters include the offset (Δx, Δy) of the bounding box center point relative to the preset anchor point, the width scaling and height scaling (log(w), log(h)), and the bounding box rotation angle θ.

[0046] Then, the raw prediction output by the detection head undergoes the following post-processing steps to obtain the final result: Decoding: The offset, width scaling, height scaling, and rotation angle predicted by the regression branch are applied to the corresponding preset anchor frame. Through geometric transformation, the initial rotated rectangle of the UFO target in the model input size image coordinate system is decoded, which is represented as (center x, center y, width w, height h, angle θ).

[0047] Filtering: Based on the confidence scores output by the classification branch, a confidence threshold is applied to filter out the preliminary rotated rectangles with low confidence, and the remaining preliminary rotated rectangles are determined as candidate rotated rectangles.

[0048] Deduplication: A non-maximum suppression algorithm for rotating bounding boxes is used to process candidate rotating bounding boxes. This algorithm calculates the intersection-union ratio between any two candidate rotating bounding boxes and iteratively retains the predicted box with the highest confidence and an overlap with other high-confidence boxes below a preset threshold to eliminate duplicate detection of the same target, ultimately obtaining a unique target rotating bounding box.

[0049] Coordinate mapping: The target rotating rectangles that have been filtered and deduplicated are linearly mapped back to the coordinate system of the original high-resolution image (3840×2160 pixels) to obtain the precise position coordinates of the UFO in the image at the moment of capture.

[0050] Output: The deep learning neural network model ultimately outputs the position information of the flying saucer in the model coordinate system as the target's spatial location. This information serves as the basis for hit determination and trajectory analysis. The entire inference process, from single-frame image input to position coordinate output, takes less than 200 milliseconds on the embedded hardware platform after model lightweighting and operator optimization, meeting real-time requirements.

[0051] In another embodiment, in step S30, determining the hit result based on the movement trajectory data and the target spatial location includes, but is not limited to, the following steps: S321, determine the target hit range based on the currently set target difficulty. The control analysis and processing module has multiple preset difficulties, and the preset difficulty is inversely proportional to the corresponding preset hit range. S322, determine the hit result based on the target hit range and the target spatial location.

[0052] It should be noted that in this embodiment, different preset difficulty levels can be set in the control analysis and processing module, such as hard, normal, and easy levels. Each preset difficulty level corresponds to a different preset hit range. For example, the preset hit range can be a circular area with the center of the image as the center. The radius of the circular area for the hard level is 150 pixels (corresponding to a diameter of 300 pixels); the radius of the circular area for the normal level is 250 pixels (corresponding to a diameter of 500 pixels); and the radius of the circular area for the easy level is 350 pixels (corresponding to a diameter of 700 pixels).

[0053] It should be noted that in this embodiment, the target hit range is obtained by expanding the image center. The coordinates of the center point of the UFO in the image coordinate system at the time of firing are determined to be within the area corresponding to the target hit range. If its center point is located within the circular area, the shot is determined to be a hit.

[0054] In another embodiment, after step S30 is performed, the following steps are included, but are not limited to: S41, after receiving the target launch signal, start timing and determine the moment when the flying saucer enters the target's hit range as the lock-on moment; S42, the follow-up time is determined based on the time interval between the locking moment and the firing moment; S43 generates shooting scores based on hit results and determines training data analysis results based on follow-up time and hit results; The S44, through the control analysis and processing module, sends shooting scores and training data analysis results to the tablet score recording module for real-time display.

[0055] It should be noted that, in order to determine the training data analysis results, this embodiment records and analyzes the timestamps of three key action nodes during this shooting process to evaluate the shooting skill level: Lock-on moment: The timer starts counting down from when the target launcher initiates the launch of the flying saucer to the moment when the flying saucer's real-time position coordinates first consistently enter the target's hit range. This moment signifies the shooter's first successful and stable alignment of the aiming baseline with the target.

[0056] Firing moment: defined as the time from the start of the target launcher launching the flying saucer to the moment when the trigger unit is triggered to generate a trigger signal.

[0057] Follow-through time: Defined as the time interval from the moment of lock-on to the moment of firing, i.e., follow-through time = firing time - lock-on time. This parameter is used to quantify the continuous tracking and fine-tuning time required for the shooter to complete the precise aiming and firing actions after locking onto the target.

[0058] It should be noted that the control analysis and processing module uses the follow time to form a quantitative evaluation index of the shooter's aiming stability, firing timing selection, and dynamic tracking ability. The specific evaluation method can be set according to actual needs, and will not be limited here.

[0059] like Figure 4 As shown, Figure 4 This is a structural diagram of a live-target clay pigeon gun simulation shooting system based on light-sensing tracking, according to one embodiment of the present invention. The present invention also provides a live-target clay pigeon gun simulation shooting system based on light-sensing tracking, comprising: The processor 401 can be implemented using a general-purpose central processing unit (CPU), microprocessor, application specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 402 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 402 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 402 and is called and executed by the processor 401 to execute the light-tracking-based simulated firing method for a clay pig gun according to the embodiments of this application. Input / output interface 403 is used to implement information input and output; The communication interface 404 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 405 transmits information between various components of the device (e.g., processor 401, memory 402, input / output interface 403, and communication interface 404); The processor 401, memory 402, input / output interface 403 and communication interface 404 are connected to each other within the device via bus 405.

[0060] This application also provides an electronic device, including the light-tracking-based live-target skeet gun simulation shooting system described above.

[0061] This application embodiment also provides a storage medium, which is a computer-readable storage medium, storing a computer program. When the computer program is executed by a processor, it implements the above-described method for simulating firing a live-target flying saucer based on light-sensing tracking.

[0062] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof. The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separate, and may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0063] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically include computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0064] The above provides a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of the present invention.

Claims

1. A live-target skeet gun simulation shooting system based on optical tracking, characterized in that, include: Multiple target launchers, responding to target launch signals, launch flying saucers; A light-sensitive gun has a muzzle and a trigger. A camera, a trigger unit, and a processing unit are arranged on the upper side of the gun body. The camera and the processing unit are electrically connected, and the trigger unit is electrically connected to the trigger and the processing unit. The shooting direction of the camera is the same as the direction of the muzzle. The central control and dispatch module is used to send the target launching signal to at least one of the target launching machines and the light-sensing gun; A control analysis and processing module is communicatively connected to the processing unit, and the control analysis and processing module is pre-configured with a deep learning network model. The processing unit responds to the target throwing signal to start the camera to acquire a real-time video stream and send it to the control analysis and processing module, and responds to the trigger signal generated by the trigger unit to control the camera to stop shooting; The control analysis and processing module determines the movement trajectory data of the light-sensing gun based on the real-time video stream, determines the firing time based on the trigger signal, acquires a visual scene image from the real-time video stream based on the firing time and inputs it into the deep learning network model, determines the target spatial position of the flying saucer based on the visual scene image through the deep learning network model, and determines the hit result based on the movement trajectory data and the target spatial position.

2. The live-target skeet gun simulation shooting system based on light-sensing tracking according to claim 1, characterized in that, The light-sensing gun is also equipped with a loading unit, a power supply unit, an audio unit, and a communication unit. The processing unit is electrically connected to the loading unit, the power supply unit, the audio unit, and the communication unit, respectively. The communication unit is communicatively connected to the control analysis and processing module.

3. The live-target skeet gun simulation shooting system based on light-sensing tracking according to claim 2, characterized in that, Also includes: The tablet performance recording module is communicatively connected to the control analysis and processing module. The tablet performance recording module is used to display the shooting scores and training data analysis results of each group in real time. The shooting scores and training data analysis results are generated by the control analysis and processing module. The network communication module includes a router and a Bluetooth device. The central control and scheduling module, the tablet score recording module, and the light sensor are connected to each other through the network communication module.

4. A method for simulating live-fire clay pigeon shooting based on optical tracking, characterized in that, The method, applied to the optically-guided live-fire clay pigeon gun simulation shooting system as described in any one of claims 1 to 3, comprises: The central control and dispatch module sends a target launching signal to at least one target launching machine and a light sensor gun, wherein the target launching machine launches the flying saucer in response to the target launching signal, and the light sensor gun activates the camera to acquire a real-time video stream in response to the target launching signal; The control analysis and processing module determines the movement trajectory data of the light sensor gun based on the real-time video stream, determines the firing time based on the trigger signal of the light sensor gun, and acquires a visual scene image from the real-time video stream based on the firing time and inputs it into the deep learning network model. The deep learning network model determines the target spatial location of the UFO based on the visual scene image, and determines the hit result based on the movement trajectory data and the target spatial location.

5. The method for simulating live-target clay pigeon shooting based on light-sensing tracking according to claim 4, characterized in that, The deep learning network model includes a backbone network, a neck network, and a detection head. The deep learning network model determines the target spatial location of the UFO based on the visual scene image, including: The visual scene image is input into the backbone network, and multiple semantic feature maps at different levels are extracted through the backbone network. The semantic feature map is upsampled and fused using the neck network to obtain multiple enhanced feature maps at different levels; Each of the enhanced feature maps is input into the detection head to determine its corresponding preset anchor point position and corresponding confidence score, and the target spatial position is determined based on the preset anchor point position.

6. The method for simulating live-target clay pigeon shooting based on light-sensing tracking according to claim 5, characterized in that, Determining the target spatial location based on the preset anchor point position includes: Determine the geometric parameters of each preset anchor point position relative to the target bounding box, wherein the geometric parameters include the offset of the target bounding box relative to the preset anchor point, the width scaling, the height scaling, and the rotation angle; Based on the geometric parameters, the initial rotation rectangle of the flying saucer in the preset model coordinate system is decoded; Based on the corresponding confidence scores and preset confidence thresholds, multiple candidate rotated rectangles are determined from the multiple preliminary rotated rectangles; The target rotating rectangle is determined from multiple candidate rotating rectangles based on a nonmaximum suppression algorithm; The coordinate information of the target rotating rectangle in the model coordinate system is used to determine the target spatial position.

7. The method for simulating live-target clay pigeon shooting based on light-sensing tracking according to claim 4, characterized in that, Determining the hit result based on the movement trajectory data and the target spatial location includes: The target hit range is determined based on the currently set target difficulty. The control analysis and processing module has multiple preset difficulty levels, and the preset difficulty level is inversely proportional to the corresponding preset hit range. The hit result is determined based on the target hit range and the target spatial location.

8. The method for simulating live-target clay pigeon shooting based on light-sensing tracking according to claim 5, characterized in that, The light-tracking-based live-target clay pigeon shooting simulation system also includes a tablet-based score recording module. After determining the hit result based on the movement trajectory data and the target's spatial position, the method further includes: After receiving the target launch signal, a timer is started, and the moment when the flying saucer enters the target's hit range is determined as the lock-on moment; The following time is determined based on the time interval between the locking moment and the firing moment; A shooting score is generated based on the hit result, and the training data analysis result is determined based on the follow-up time and the hit result; The shooting scores and training data analysis results are sent to the tablet score recording module for real-time display via the control analysis and processing module.

9. A live-fire target skeet gun simulation device based on light-sensing tracking, comprising at least one control processor and a memory for communicatively connecting to the at least one control processor; the memory stores instructions executable by the at least one control processor, the instructions being executed by the at least one control processor to enable the at least one control processor to perform the live-fire target skeet gun simulation method based on light-sensing tracking as described in any one of claims 4 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions for causing a computer to perform the light-sensing tracking-based simulated firing method for a live-target skeet gun as described in any one of claims 4 to 8.

Images