Shooting effect evaluation method based on human body analysis

Abstract

The invention discloses a shooting evaluation effect method based on human body analysis, and relates to the technical field of computer vision. According to the method, the one-sidedness of single index evaluation is avoided by calculating the ratio of the hit times in the total training shooting times, the continuous hit times, the highest uninterrupted hit times and the shooting level value after conversion of the hit performance scores of different parts, and the ability of a shooting player can be evaluated more comprehensively and stereoscopically; meanwhile, by setting corresponding pass standards and influence weight factors, various evaluation indexes are quantified, and the shooting comprehensive performance index is obtained through weighted calculation, so that the evaluation result is more objective and accurate, the performance of players can be conveniently compared and sorted, and the training strategy can be adjusted in a targeted mode.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and in particular to a shooting effect evaluation method based on human body analysis. Background Art

[0002] In the modern military and police fields, shooting training and strike evaluation are important means to improve shooting accuracy and combat capabilities. With the development of image processing technology, computer vision and deep learning technology, automated shooting evaluation based on human body analysis has become possible. Human body analysis technology can accurately identify and segment targets by analyzing human body parts in images, thereby improving the accuracy and real-time performance of shooting evaluation.

[0003] However, the existing technical solutions still have the following deficiencies: The aiming image and human body analysis algorithm during shooting were not effectively combined, resulting in the inability to form a comprehensive and efficient method for evaluating the effect of the strike; The judging methods in the past were inefficient and involved too much human intervention. Shooting judging in the past was often based on a single indicator such as the number of hits, which could not fully reflect the true level of the shooting athletes.

[0004] Therefore, a shooting effect evaluation method based on human body analysis is proposed. Summary of the invention

[0005] The purpose of the present invention is to solve the problems pointed out in the background technology and to propose a shooting evaluation method based on human body analysis.

[0006] The purpose of the present invention can be achieved by the following technical solution: A shooting evaluation method based on human body analysis, comprising: Image processing: At the moment of shooting, the shooting image is collected and preprocessed, and the preprocessed shooting image is segmented to segment the various parts of the human body, including the head, torso, hands and feet; Shooting evaluation: extract the coordinates of the aiming point during shooting, that is, the relative position of the aiming point in the image, and judge whether the aiming point hits the expected part of the target based on the positional relationship between the coordinates of the aiming point and the segmented parts; Signal generation: If the aiming point falls within the internal distribution area corresponding to the expected part of the target, it is judged as hitting the target, and a "hit the enemy" signal is fed back; if the aiming point does not hit any part of the target, it is judged as not hitting the target, and a "missed enemy" signal is fed back; Shooting effect: Obtain the shooting data of each shooter in the current training process; the shooting data includes the total number of training shootings of each shooter, the number of feedback signals of "hitting the enemy" and the number of feedback signals of "missing the enemy"; conduct a comprehensive evaluation of the shooting data of each shooter in the current training process to determine the shooting performance level of the shooter in the current training process; the shooting performance level includes poor shooting level, passing shooting level and excellent shooting level.

[0007] As a preferred embodiment of the present invention, the pre-processed shooting image is segmented to separate various parts of the human body, including the head, torso, hands and feet, specifically: Construct a human body analysis algorithm framework, which adopts a deep learning model combined with a convolutional neural network structure. The framework includes an image input layer, a feature extraction layer, a classification layer, and an output layer; This framework can be represented as the following neural network model: ; Where: is the input image, i.e. the image aimed at at the moment of shooting, are the trainable parameters of the neural network, and the neural network optimizes these parameters through training. The output image is the analysis result of each part of the human body obtained after the neural network analysis; Collect human body image data in shooting scenes as shooting images and manually annotate them. The image data includes target human body images at different shooting angles, environmental backgrounds, and different postures. The annotation includes accurately marking various parts of the human body and generating a training data set based on the annotation results. The label of each image obtained by manual annotation is , then the label of each image can be expressed as: ; Where: Indicates the head, Represents the torso, Indicates hand, Indicates the legs; Training the deep learning network using the labeled training data set, the training process includes optimizing the network weights using a back-propagation algorithm and then adjusting the network parameters by minimizing a loss function; The loss function uses the cross entropy loss function: ; Where: is the true label, For the model prediction results, is the number of samples, the cross entropy loss function is used to measure the difference between the output and the label, and the loss function is minimized by the gradient descent method. , gradually adjust the network parameters ; Verifying the trained model, wherein the verification process is performed using an independent verification set, during which the model is also error analyzed and model parameters or training strategies are adjusted accordingly; The trained model is quantized and pruned. The quantization pruning process includes weight pruning and quantization of the neural network, and the floating precision calculation is converted into low precision calculation through quantization technology. The optimized model is deployed to the neural network processing unit NPU in the edge device, and the NPU unit is embedded in the gun sight.

[0008] As a preferred embodiment of the present invention, whether the aiming point hits the expected part of the target is determined as follows: The geometric model of the target area is established. The boundary of each part is described by a series of coordinate points, usually the contour of the part. The boundary of each part is set to , which is represented as a closed polygonal area or a set of curves; By judging the aiming point Is it located at the boundary of each part? Internally, to determine whether it hits; for each part Determine whether the aiming point is within the area of ​​the part, using the formula ; Where: It is a hit function, which indicates whether the aiming point is inside the target part. If the aiming point is within the boundary of the part, it returns 1 to indicate a hit, otherwise it returns 0 to indicate a miss.

[0009] As a preferred embodiment of the present invention, a comprehensive evaluation is performed on the shooting data of each shooting player in the current training process, specifically: Extract the number of feedback "hit the enemy" signals from the shooting data of each shooter as the number of hits, calculate the proportion of the number of hits in the total number of training shootings, that is, calculate it by the number of hits / total number of training shootings, and mark the calculated proportion as the hit proportion; A reference number for determining the number of consecutive hits is set; the number of hits of each shooter is identified based on the set reference number, and if there are uninterrupted hits in each hit of a certain shooter, and the number of uninterrupted hits reaches the set reference number, it is determined to be a continuous hit behavior; the number of consecutive hits of each shooter is counted as a stable evaluation value; the highest number of uninterrupted hits in each group of consecutive hits of each shooter is extracted as a limit evaluation value; The number of hits of each shooter is analyzed and classified according to the target parts hit, namely head, torso, hands and feet; each target part is set to correspond to a hit performance score; The number of hits of each shooter is converted into a hit performance score according to the target part hit. After the conversion is completed, the hit performance scores of each group of shooters are accumulated to obtain the shooting level value of each shooter.

[0010] As a preferred embodiment of the present invention, a comprehensive evaluation of the shooting data of each shooting player in the current training process also includes: Set the passing hit percentage, passing stability evaluation value, normal limit evaluation value and normal shooting level value corresponding to the hit percentage, stability evaluation value, limit evaluation value and shooting level value in the current training process, and mark the passing hit percentage, passing stability evaluation value, normal limit evaluation value and normal shooting level value as Ke1, Ke2, Ke3 and Ke4 respectively; Extract the hit percentage, stable evaluation value, limit evaluation value and shooting level value of each shooter and mark them as kg1, kg2, kg3 and kg4 respectively; and substitute them into the formula Perform weighted calculation to obtain the comprehensive shooting performance index Sgh of each shooter in the current training process; , , as well as They are the influence weight factors corresponding to the hit proportion kg1, stability evaluation value kg2, limit evaluation value kg3 and shooting level value kg4 respectively, e is a preset natural constant, and e>1.

[0011] As a preferred embodiment of the present invention, the shooting performance level of the shooting player in the current training process is determined as follows: The three groups of index value ranges corresponding to the preset shooting comprehensive performance index Sgh are set, and each group of index value ranges is set to correspond to a shooting performance level. The shooting comprehensive performance index Sgh of each shooter in the current training process is matched with the preset three groups of index value ranges, so as to determine the shooting performance level of the shooter in the current training process.

[0012] As a preferred implementation of the present invention, a recent shooting report of each shooting contestant is generated, specifically: Taking the time point of the current training process as the starting point, extract the shooting comprehensive performance index of each shooter in the V groups of historical training before the starting point as the recent shooting status evaluation data of each shooter; where V>5; The standard deviation formula is used to calculate the comprehensive shooting performance index of each group of each shooter, so as to obtain the discrete value of each shooter's shooting performance in the recent period; At the same time, the shooting comprehensive performance indexes of each shooting contestant are arranged from large to small, and after removing the highest and lowest values, the average of the remaining shooting comprehensive performance indexes is calculated, and then matched with the preset three groups of index value ranges to determine the shooting performance level of each shooting contestant in the near future; A reference discrete value corresponding to the shooting performance discrete value is set. If a shooting athlete's recent shooting performance discrete value is higher than the set reference discrete value, the shooting athlete's state higher than the reference discrete value is marked as unstable, and the shooting athlete's state lower than the reference discrete value is marked as stable; Based on the recent shooting performance level and status of each shooter, construct the shooting performance data pair of each shooter, i.e. (shooting performance level / status); The shooting performance level and shooting performance data of each shooter in the current training process are input into a pre-built data template to obtain a shooting report of each shooter.

[0013] Compared with the prior art, the present invention has the following beneficial effects: The present invention avoids the one-sidedness of single index evaluation by calculating the proportion of hit times in the total number of training shooting times, the number of consecutive hits, the highest number of uninterrupted hits and the shooting level value after the hit performance scores of different parts are converted, and can evaluate the ability of shooting players more comprehensively and three-dimensionally; at the same time, by setting corresponding passing standards and influencing weight factors, various evaluation indicators are quantified, and the comprehensive shooting performance index is obtained by weighted calculation, so that the evaluation result is more objective and accurate, which is convenient for comparing and ranking the performance of players and also helps to adjust the training strategy in a targeted manner; The present invention presets three groups of index value ranges corresponding to different shooting performance levels, matches the shooting comprehensive performance index with them, and can more finely divide the shooting performance levels of the players, so that the players and related personnel have a clearer understanding of their performance positioning in the current training, which helps to adjust the training strategy in a targeted manner; The present invention extracts the comprehensive shooting performance index of multiple groups of historical training processes before the starting point as the recent shooting status evaluation data, and comprehensively analyzes the performance of the players from the time dimension. Compared with only focusing on the current training data, it can more comprehensively reflect the real level and development trend of the players, and avoid the influence of accidental factors of a single training on the judgment of the players' abilities. The present invention uses a deep learning model to perform precise pixel-level segmentation of various parts of the human body, and can provide accurate impact effect evaluation. Through automated image analysis and impact evaluation, the hit status can be returned in real time, which greatly improves the evaluation efficiency and reduces manual intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0015] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0016] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] See also Figure 1 As shown, a shooting evaluation method based on human body analysis includes: Image processing: At the moment of shooting, the shooting image is collected and preprocessed; the preprocessing includes image cropping, denoising and enhancement, etc. Use the human body analysis algorithm to segment the preprocessed shooting images and segment the various parts of the human body, including the head, torso, hands and feet; It should be noted that the device used to collect images is specifically a digital intelligent individual combat gun sight; The preprocessed shooting images are segmented to separate the various parts of the human body, including the head, torso, hands and feet. Specifically: Construct a human body analysis algorithm framework, which uses a deep learning model combined with a convolutional neural network structure to improve the accuracy of segmentation of various parts of the human body through a network hierarchical design. The framework includes an image input layer, a feature extraction layer, a classification layer, and an output layer; This framework can be represented as the following neural network model: ; Where: is the input image, i.e. the image aimed at at the moment of shooting, are the trainable parameters of the neural network, and the neural network optimizes these parameters through training. The output image is the analysis result of each part of the human body obtained after the neural network analysis; Collect human body image data in shooting scenes as shooting images and manually annotate them. The image data includes target human body images at different shooting angles, environmental backgrounds, and different postures. The annotation includes accurately marking various parts of the human body and generating a training data set based on the annotation results. The label of each image obtained by manual annotation is , then the label of each image can be expressed as: ; Where: Indicates the head, Represents the torso, Indicates hand, Indicates the legs; The deep learning network is trained using the labeled training data set, wherein the training process includes optimizing the network weights using a back-propagation algorithm, and then adjusting the network parameters by minimizing a loss function (such as a cross entropy loss function) to improve the accuracy of the model in segmenting human body parts; The loss function uses the cross entropy loss function: ; Where: is the true label, For the model prediction results, is the number of samples, the cross entropy loss function is used to measure the difference between the output and the label, and the loss function is minimized by the gradient descent method. , gradually adjust the network parameters ; The trained model is verified by using an independent verification set, and evaluation indicators such as accuracy, recall, and F1 value on the verification set are calculated to ensure the generalization ability of the model in different environments and postures. During the verification process, the model is also analyzed for errors, and model parameters or training strategies are adjusted in a targeted manner; Quantize and prune the trained model. The quantization pruning process includes weight pruning and quantization of the neural network to reduce unnecessary computing resource consumption. Quantization technology is used to convert floating precision calculations into low precision calculations to reduce model storage and computing requirements, ensuring that the model can run efficiently on embedded devices. The optimized model is deployed to the neural network processing unit (NPU) in the edge device. The NPU unit is embedded in the gun sight, and fast human body part analysis and aiming point judgment are achieved through edge computing, ensuring that the method can be executed in real time and efficiently, and providing accurate strike judgment results at the moment of shooting. Shooting evaluation: extract the coordinates of the aiming point during shooting, that is, the relative position of the aiming point in the image, and judge whether the aiming point hits the expected part of the target based on the positional relationship between the coordinates of the aiming point and the segmented parts; It should be noted that the aiming point coordinates It is extracted from the image sent back by the rifle scope through an image processing algorithm and represents the shooter’s aiming position at the moment of shooting; Determine whether the aiming point hits the expected part of the target, specifically: The geometric model of the target area is established. The boundary of each part is described by a series of coordinate points, usually the contour of the part. The boundary of each part is set to , which is represented as a closed polygonal area or a set of curves; By judging the aiming point Is it located at the boundary of each part? Internally, to determine whether it hits; for each part Determine whether the aiming point is within the area of ​​the part, using the formula ; Where: is the hit function, indicating whether the aiming point is inside the target part. If the aiming point is within the boundary of the part, it returns 1 to indicate a hit, otherwise it returns 0 to indicate a miss; Signal generation: If the aiming point falls within the internal distribution area corresponding to the expected part of the target, it is judged as hitting the target, and a "hit the enemy" signal is fed back; if the aiming point does not hit any part of the target, it is judged as not hitting the target, and a "missed enemy" signal is fed back; It should be noted that after the shooting effect is evaluated, a hit or miss judgment result is generated based on the relationship between the aiming point and various parts of the target. The generated result is transmitted to the mobile terminal through a predetermined wireless communication protocol, and the mobile terminal provides real-time feedback after receiving the result.

[0018] Shooting effect: Obtain the shooting data of each shooter in the current training process; the shooting data includes the total number of training shootings of each shooter, the number of feedback signals of "hitting the enemy" and the number of feedback signals of "missing the enemy"; conduct a comprehensive evaluation of the shooting data of each shooter in the current training process to determine the shooting performance level of the shooter in the current training process; the shooting performance level includes poor shooting level, passing shooting level and excellent shooting level; Specifically: Extract the number of feedback "hit the enemy" signals from the shooting data of each shooter as the number of hits, calculate the proportion of the number of hits in the total number of training shootings, that is, calculate it by the number of hits / total number of training shootings, and mark the calculated proportion as the hit proportion; Set a reference number for determining the number of consecutive hits; set it to 3 times; identify the number of hits of each shooter based on the set reference number, and if there are uninterrupted hits in each hit of a certain shooter, and the number of uninterrupted hits reaches the set reference number, it is determined to be a continuous hit behavior; count the number of consecutive hits of each shooter as a stable evaluation value; extract the highest number of uninterrupted hits in each group of consecutive hits of each shooter as a limit evaluation value; The number of hits of each shooter was analyzed and classified according to the target parts hit, namely head, torso, hands and feet; Set a hit performance score for each target part; the hit performance score range is set to 1-5, and the hit performance score of the head> the hit performance score of the torso> the hit performance score of the feet> the hit performance score of the hands; The number of hits of each shooter is converted into a hit performance score according to the target parts hit, and after the conversion is completed, the hit performance scores of each group of shooters are accumulated to obtain the shooting level value of each shooter; Set the passing hit percentage, passing stability evaluation value, normal limit evaluation value and normal shooting level value corresponding to the hit percentage, stability evaluation value, limit evaluation value and shooting level value in the current training process, and mark the passing hit percentage, passing stability evaluation value, normal limit evaluation value and normal shooting level value as Ke1, Ke2, Ke3 and Ke4 respectively; Extract the hit percentage, stable evaluation value, limit evaluation value and shooting level value of each shooter and mark them as kg1, kg2, kg3 and kg4 respectively; and substitute them into the formula Perform weighted calculation to obtain the comprehensive shooting performance index Sgh of each shooter in the current training process; , , as well as They are the impact weight factors corresponding to the hit ratio kg1, the stability evaluation value kg2, the limit evaluation value kg3 and the shooting level value kg4, e is a preset natural constant, and e>1; the specific value of e can be set to 1.257; It should be noted that the above not only takes into account the simple hit percentage, but also includes multiple factors such as the number of consecutive hits (stable evaluation value), the highest number of uninterrupted hits (limit evaluation value) and the shooting level value after the hit performance points of different parts are converted. This avoids the one-sidedness of evaluating the shooting level based on only a single indicator, and can comprehensively and three-dimensionally reflect the real level of the shooting athletes; By setting a reference number of consecutive hits to calculate the stability evaluation value, the shooter's ability to keep hitting during continuous shooting can be measured, highlighting the importance of shooting stability; the hit parts are classified and different performance points are set to calculate the shooting level value, which reflects the requirements for shooting accuracy and guides the shooters to pursue higher shooting quality; Quantify each evaluation indicator, set corresponding passing standards and influencing weight factors, and calculate the comprehensive shooting performance index through weighted calculation, so that the evaluation results are more objective and accurate, and it is convenient to compare and rank the performance of each shooting athlete.

[0019] Three groups of index value ranges corresponding to the shooting comprehensive performance index Sgh are preset, and each group of index value ranges is set to correspond to a shooting performance level. The shooting comprehensive performance index Sgh of each shooter in the current training process is matched with the preset three groups of index value ranges, so as to determine the shooting performance level of the shooter in the current training process; the higher the shooting comprehensive performance index of the shooter, the higher the possibility of matching the excellent shooting level; It should be noted that presetting three groups of index value ranges and corresponding to different shooting performance levels, and matching the shooting comprehensive performance index with them, can more finely divide the shooting performance levels of the players, so that the players and relevant personnel have a clearer understanding of their performance positioning in the current training, which is helpful to adjust the training strategy in a targeted manner; In addition, taking the time point of the current training process as the starting point, extract the shooting comprehensive performance index of each shooter in the V groups of historical training before the starting point as the recent shooting status evaluation data of each shooter; where V>5, the specific extraction number is set by the technical staff; The standard deviation formula is used to calculate the comprehensive shooting performance index of each group of each shooter, so as to obtain the discrete value of each shooter's shooting performance in the recent period; The lower the shooting performance discrete value, the more stable the shooting state of the shooter in the recent period. At the same time, the shooting comprehensive performance indexes of each shooting contestant are arranged from large to small, and after removing the highest and lowest values, the average of the remaining shooting comprehensive performance indexes is calculated, and then matched with the preset three groups of index value ranges to determine the shooting performance level of each shooting contestant in the near future; A reference discrete value corresponding to the shooting performance discrete value is set. If a shooting athlete's recent shooting performance discrete value is higher than the set reference discrete value, the shooting athlete's state higher than the reference discrete value is marked as unstable, and the shooting athlete's state lower than the reference discrete value is marked as stable; Based on the recent shooting performance level and status of each shooter, construct the shooting performance data pair of each shooter, i.e. (shooting performance level / status); The shooting performance level and shooting performance data of each shooter in the current training process are input into a pre-built data template to obtain a shooting report of each shooter; It should be noted that extracting the comprehensive shooting performance index of multiple groups of historical training processes before the starting point as the recent shooting status evaluation data, and comprehensively analyzing the player's performance from the time dimension, can more comprehensively reflect the player's true level and development trend, and avoid the influence of accidental factors of a single training on the judgment of the player's ability.

[0020] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A shooting evaluation method based on human body analysis, characterized in that: include: Image processing: At the moment of shooting, the shooting image is collected and preprocessed, and the preprocessed shooting image is segmented to segment the various parts of the human body, including the head, torso, hands and feet; Shooting evaluation: extract the coordinates of the aiming point during shooting, that is, the relative position of the aiming point in the image, and judge whether the aiming point hits the expected part of the target based on the positional relationship between the coordinates of the aiming point and the segmented parts; Signal generation: If the aiming point falls within the internal distribution area corresponding to the expected part of the target, it is judged as hitting the target, and a "hit the enemy" signal is fed back; if the aiming point does not hit any part of the target, it is judged as not hitting the target, and a "missed enemy" signal is fed back; Shooting effect: Obtain the shooting data of each shooter in the current training process; the shooting data includes the total number of training shootings of each shooter, the number of feedback signals of "hitting the enemy" and the number of feedback signals of "missing the enemy"; conduct a comprehensive evaluation of the shooting data of each shooter in the current training process, so as to determine the shooting performance level of the shooter in the current training process; The shooting performance levels include poor shooting level, passing shooting level and excellent shooting level.

2. A shooting evaluation method based on human body analysis according to claim 1, characterized in that: The preprocessed shooting images are segmented to separate the various parts of the human body, including the head, torso, hands and feet. Specifically: Construct a human body analysis algorithm framework, which adopts a deep learning model combined with a convolutional neural network structure. The framework includes an image input layer, a feature extraction layer, a classification layer, and an output layer; This framework can be represented as the following neural network model: ； Where: is the input image, i.e. the image aimed at at the moment of shooting, are the trainable parameters of the neural network, and the neural network optimizes these parameters through training. The output image is the analysis result of each part of the human body obtained after the neural network analysis; Collect human body image data in shooting scenes as shooting images and manually annotate them. The image data includes target human body images at different shooting angles, environmental backgrounds, and different postures. The annotation includes accurately marking various parts of the human body and generating a training data set based on the annotation results. The label of each image obtained by manual annotation is , then the label of each image can be expressed as: ； Where: Indicates the head, Represents the torso, Indicates hand, Indicates the legs; Training the deep learning network using the labeled training data set, the training process includes optimizing the network weights using a back-propagation algorithm and then adjusting the network parameters by minimizing a loss function; The loss function uses the cross entropy loss function: ； Where: is the true label, For the model prediction results, is the number of samples, the cross entropy loss function is used to measure the difference between the output and the label, and the loss function is minimized by the gradient descent method. , gradually adjust the network parameters ; Verifying the trained model, wherein the verification process is performed using an independent verification set, during which the model is also error analyzed and model parameters or training strategies are adjusted accordingly; The trained model is quantized and pruned. The quantization pruning process includes weight pruning and quantization of the neural network, and the floating precision calculation is converted into low precision calculation through quantization technology. The optimized model is deployed to the neural network processing unit NPU in the edge device, and the NPU unit is embedded in the gun sight.

3. The shooting effect evaluation method based on human body analysis according to claim 2 is characterized in that: Determine whether the aiming point hits the expected part of the target, specifically: The geometric model of the target area is established. The boundary of each part is described by a series of coordinate points, usually the contour of the part. The boundary of each part is set to , which is represented as a closed polygonal area or a set of curves; By judging the aiming point Is it located at the boundary of each part? Internally, to determine whether it hits; for each part Determine whether the aiming point is within the area of ​​the part, using the formula ; Where: It is a hit function, which indicates whether the aiming point is inside the target part. If the aiming point is within the boundary of the part, it returns 1 to indicate a hit, otherwise it returns 0 to indicate a miss.

4. The shooting effect evaluation method based on human body analysis according to claim 3 is characterized in that: Comprehensively evaluate the shooting data of each shooter during the current training process, specifically: Extract the number of feedback "hit the enemy" signals from the shooting data of each shooter as the number of hits, calculate the proportion of the number of hits in the total number of training shootings, that is, calculate it by the number of hits / total number of training shootings, and mark the calculated proportion as the hit proportion; A reference number for determining the number of consecutive hits is set; the number of hits of each shooter is identified based on the set reference number, and if there are uninterrupted hits in each hit of a certain shooter, and the number of uninterrupted hits reaches the set reference number, it is determined to be a continuous hit behavior; the number of consecutive hits of each shooter is counted as a stable evaluation value; the highest number of uninterrupted hits in each group of consecutive hits of each shooter is extracted as a limit evaluation value; The number of hits of each shooter is analyzed and classified according to the target parts hit, namely head, torso, hands and feet; each target part is set to correspond to a hit performance score; The number of hits of each shooter is converted into a hit performance score according to the target part hit. After the conversion is completed, the hit performance scores of each group of shooters are accumulated to obtain the shooting level value of each shooter.

5. The shooting effect evaluation method based on human body analysis according to claim 4 is characterized in that: Comprehensive evaluation of each shooter's shooting data during the current training process, including: Set the passing hit percentage, passing stability evaluation value, normal limit evaluation value and normal shooting level value corresponding to the hit percentage, stability evaluation value, limit evaluation value and shooting level value in the current training process, and mark the passing hit percentage, passing stability evaluation value, normal limit evaluation value and normal shooting level value as Ke1, Ke2, Ke3 and Ke4 respectively; Extract the hit percentage, stable evaluation value, limit evaluation value and shooting level value of each shooter and mark them as kg1, kg2, kg3 and kg4 respectively; and substitute them into the formula Perform weighted calculation to obtain the comprehensive shooting performance index Sgh of each shooter in the current training process; , , as well as They are the influence weight factors corresponding to the hit proportion kg1, stability evaluation value kg2, limit evaluation value kg3 and shooting level value kg4 respectively, e is a preset natural constant, and e>

1.

6. The shooting effect evaluation method based on human body analysis according to claim 5, characterized in that: Determine the shooting performance level of the shooter during the current training process, specifically: The three groups of index value ranges corresponding to the preset shooting comprehensive performance index Sgh are set, and each group of index value ranges is set to correspond to a shooting performance level. The shooting comprehensive performance index Sgh of each shooter in the current training process is matched with the preset three groups of index value ranges, so as to determine the shooting performance level of the shooter in the current training process.

7. The shooting effect evaluation method based on human body analysis according to claim 6, characterized in that: Generate recent shooting reports for each shooter, specifically: Taking the time point of the current training process as the starting point, extract the shooting comprehensive performance index of each shooter in the V groups of historical training before the starting point as the recent shooting status evaluation data of each shooter; where V>5; The standard deviation formula is used to calculate the comprehensive shooting performance index of each group of each shooter, so as to obtain the discrete value of each shooter's shooting performance in the recent period; At the same time, the shooting comprehensive performance indexes of each shooting contestant are arranged from large to small, and after removing the highest and lowest values, the average of the remaining shooting comprehensive performance indexes is calculated, and then matched with the preset three groups of index value ranges to determine the shooting performance level of each shooting contestant in the near future; A reference discrete value corresponding to the shooting performance discrete value is set. If a shooting athlete's recent shooting performance discrete value is higher than the set reference discrete value, the shooting athlete's state higher than the reference discrete value is marked as unstable, and the shooting athlete's state lower than the reference discrete value is marked as stable; Based on the recent shooting performance level and status of each shooter, construct the shooting performance data pair of each shooter, i.e. (shooting performance level / status); The shooting performance level and shooting performance data of each shooter in the current training process are input into a pre-built data template to obtain a shooting report of each shooter.

Images