Improved 4S store passenger flow statistical algorithm based on pedestrian re-identification

Abstract

The invention relates to the technical field of computer vision, in particular to an improved 4S store passenger flow statistical algorithm based on pedestrian re-identification, which comprises the following steps of acquiring video data of a 4S store entrance in real time based on image acquisition equipment, performing standardization processing on the video data in combination with an input requirement of video processing, and generating a video standardization processing result. According to the method, the image quality and the data processing continuity under the condition of high passenger flow are ensured by collecting the video data in real time and performing standardization processing, the accuracy of target identification and tracking is improved by combining the adjustment of the size and the position of the target detection frame, the motion path of the target is analyzed, and the mark information is added; the trend of a plurality of customers is accurately tracked and compared with feature vectors stored in a database, repeated targets are effectively distinguished, passenger flow statistics is carried out, in-out behavior analysis is carried out in combination with store layout parameters, a storefront is helped to adjust human resource allocation, and the operation efficiency and customer satisfaction are improved.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and in particular to an improved 4S store customer flow statistics algorithm based on pedestrian re-identification. Background Art

[0002] The field of computer vision technology uses artificial intelligence and deep learning to analyze image data captured by cameras and sensors, enabling machines to understand visual information, and parse image data through algorithms to identify and classify objects, scenes and activities. It performs multiple tasks such as image processing, target detection, image classification and image reconstruction without human intervention. It is applied to a variety of practical scenarios, including autonomous driving, industrial automation, medical imaging analysis, security monitoring and human-computer interaction. By improving and optimizing the accuracy and processing speed of the algorithm, the system can work stably in a variety of complex environments and conditions.

[0003] Among them, the 4S store customer flow statistics algorithm focuses on customer flow statistics for 4S car shows and service stores. The system uses target detection, multi-target tracking, and pedestrian re-identification technology in computer vision technology to accurately calculate the number of customers entering and leaving the store. By distinguishing and tracking each customer, it maintains a high accuracy rate when there are a large number of customers and frequent entries and exits. It is used in business environments that require accurate customer flow data and optimize store operations and improve customer service experience. It manages customer flow, rationally arranges personnel and resources, and improves operational efficiency and customer satisfaction.

[0004] Traditional 4S store customer flow statistics technology has shortcomings in the field of commercial customer flow statistics. It relies on simple image processing technology, which makes it susceptible to light and occlusion factors in high customer flow and complex environments. When processing large amounts of real-time data, it requires higher computing resources, which limits its application in resource-constrained environments. It lacks effective customer behavior tracking and analysis capabilities, making it difficult to provide merchants with in-depth customer behavior insights. The traffic statistics system cannot distinguish between new and old customers, making it difficult for merchants to effectively implement targeted marketing strategies, affecting the application effect of commercial users on customer flow data, limiting the flexibility and response speed of commercial operations, and reducing customer satisfaction and the economic benefits of merchants. Summary of the invention

[0005] In order to solve the technical problem that the existing technology has labor costs, equipment funds, and installation space factors, which make it difficult to meet the growing number of modern 4S stores, the embodiment of the present invention provides a 4S store customer flow statistics algorithm and device based on improved pedestrian re-identification. The technical solution is as follows: On the one hand, a 4S store customer flow counting algorithm based on improved pedestrian re-identification is provided, and the method is implemented by an SSS device. The method includes: S1: Based on the image acquisition device, the video data of the 4S store entrance is collected in real time, and the video data is standardized according to the input requirements of video processing to generate the video standardization processing results; S2: Based on the video standardization processing result, the learning speed is configured to match the dynamic changes in the video, the objects in the video data are detected, the size and position of the recognition frame are adjusted to reduce the overlap between multiple frames, and the target detection frame information is obtained; S3: Based on the target detection frame information, by analyzing the position parameters of the identified target, calculating the position change of the target between consecutive frames, identifying the motion path of the target and adding marking information, and generating movement path tracking information; S4: Based on the mobile path tracking information, feature vectors of multiple targets are extracted and compared with feature vectors stored in a database, similarity scores between feature data are calculated, and a re-identification classification result is obtained; S5: Based on the re-identification classification result, an entry detection buffer is set in combination with store layout parameters, the entry and exit behaviors of multiple targets are analyzed in real time, the entry and exit status of multiple targets are updated and recorded, and an area entry detection result is generated; S6: Based on the area entry detection results, count the pedestrians entering and exiting the store entrance, record the characteristics and identification information of multiple pedestrians, evaluate the human resource allocation required for the operation of the target store, and generate store customer flow statistics.

[0006] As a further solution of the present invention, the video standardization processing result includes image resolution adjustment information, frame rate adjustment record, and timestamp record information; the target detection frame information includes positioning recognition frame size parameters, recognition frame position information, and recognition frame positioning position information; the mobile path tracking information includes the target's motion speed data, motion direction information, and motion trajectory time series; the re-identification classification result includes the target's feature similarity score, a list of targets with confirmed identities, and feature library matching information of targets to be re-identified; the area entry detection result includes the target entry and exit status record, the target's behavior analysis in the buffer zone, and the target's spatial position change record; the store customer flow statistics data includes the counted number of pedestrians, pedestrian features and identification data sets, and customer flow change trend data.

[0007] As a further solution of the present invention, based on the image acquisition device, the video data of the 4S store entrance is collected in real time, and the video data is standardized in combination with the input requirements of video processing. The specific steps of generating the video standardization processing result are as follows: S101: Based on the image acquisition device, the focal length and viewing angle of the camera are adjusted to optimize the acquisition range, and the video data of the entrance of the 4S store is collected in real time. The time information of the acquisition device is calibrated to optimize the consistency between the video data and the actual time, and the video acquisition result is generated; S102: Based on the video acquisition result and according to the input requirements of the video processing, adjusting the resolution of multiple frames of images in the video, optimizing the image clarity, and generating a resolution-adjusted video stream; S103: adjusting the video stream based on the resolution, adjusting the frame rate of the video stream, setting the frame rate to match the dynamic scene changes of the video surveillance, and generating a video standardization processing result.

[0008] As a further solution of the present invention, based on the video normalization processing result, by configuring the learning speed to match the dynamic changes in the video, detecting the object in the video data, adjusting the size and position of the recognition frame to reduce the overlap between multiple frames, the steps of obtaining the target detection frame information are specifically as follows: S201: Based on the video normalization processing result, adjusting the learning rate of the target detection model to match the dynamic changes in the video data, optimizing the response speed of the model to various scenes, and generating a learning rate adjustment result; S202: Based on the learning rate adjustment result, detecting moving objects in the video frame, recording positions and movement information of multiple targets, and generating target position data; S203: Based on the target position data, the size and position of the recognition frame are adjusted to reduce overlap between frames, optimize recognition accuracy and target tracking capability, and generate target detection frame information.

[0009] As a further solution of the present invention, the specific formula for adjusting the learning rate of the target detection model is: in, is the learning rate adjustment function, calculating the current number of iterations The learning rate function combines the cosine annealing algorithm to optimize the dynamic adjustment of the learning rate, ensuring that the learning rate is large at the beginning of training to converge quickly, and gradually decreases near the end of training to refine the adjustment of model parameters. is the current iteration number, indicating the time point in the learning process, and is used to adjust the instantaneous value of the learning rate. is the total number of iterations, defined as the total number of preset iteration cycles in the entire training process, used for normalization ratio to ensure that the learning rate adjustment is reasonably distributed within the training cycle. is the minimum value of the learning rate, ensuring that the learning rate will not drop to an extremely low level that is invalid for model training during the reduction process, thereby maintaining the effectiveness and stability of the algorithm throughout the training process. is pi, a constant introduced in the calculation to ensure that the adjustment of the learning rate has a periodic and smooth transition to adapt to different training stages.

[0010] As a further solution of the present invention, based on the target detection frame information, by analyzing the position parameters of the identified target, calculating the position change of the target between consecutive frames, identifying the motion path of the target and adding marking information, the steps of generating the moving path tracking information are specifically as follows: S301: Analyze and identify the position coordinates of multiple targets based on the target detection frame information, compare the coordinate data between consecutive frames, analyze the moving direction and speed of the target in the video, and generate a position change analysis result; S302: Based on the position change analysis result, the moving paths of multiple targets are tracked in real time, and the moving trajectories and durations of the targets in the video sequence are recorded to generate trajectory recording results; S303: Based on the trajectory recording result, identification marks are added to multiple targets by associating with the movement trajectory and time stamp of the target to generate movement path tracking information.

[0011] As a further solution of the present invention, based on the mobile path tracking information, the steps of extracting feature vectors of multiple targets and comparing them with feature vectors stored in a database, calculating similarity scores between feature data, and obtaining re-identification classification results are specifically as follows: S401: Based on the movement path tracking information, identifying feature vectors of multiple targets according to image information, including size, geometric shape, and color information of the targets, and generating feature vector extraction results; S402: Based on the feature vector extraction result, by comparing with the identified targets in the database, calculating the similarity score between the two sets of vectors, and generating a feature matching result; S403: Based on the feature matching result, by determining the matching status of the target and the known targets in the database, classifying the matched target and updating the identity tag, identifying the re-identification status of the target, and generating a re-identification classification result.

[0012] As a further solution of the present invention, based on the re-identification classification result, the entry detection buffer is set in combination with the store layout parameters, the entry and exit behaviors of multiple targets are analyzed in real time, the entry and exit status of multiple targets are updated and recorded, and the steps of generating the area entry detection result are specifically as follows: S501: Based on the re-identification and classification results, the layout of the target 4S store is analyzed, and boundary parameters of the entrance buffer are configured, including the spatial layout of the store entrance and the direction of pedestrian flow, and a buffer parameter configuration is generated; S502: Based on the buffer parameter configuration, by real-time analysis of the video data, using the position data to determine and identify the target that crosses the buffer, recording the entry and exit status of the target, and generating a real-time entry and exit status record; S503: According to the real-time entry and exit status record, the target entry and exit data in the database is updated in real time, including the identification code and entry and exit time information for identifying the target, and the area entry detection result is generated.

[0013] As a further solution of the present invention, based on the area entry detection result, the pedestrians entering and exiting the store entrance are counted, the characteristics and identification information of multiple pedestrians are recorded, the manpower configuration required for the operation of the target store is evaluated, and the steps of generating store passenger flow statistics are specifically as follows: S601: Based on the area entry detection result, calculate the pedestrian flow at the store entrance, including the number of people entering and leaving the store, and generate a pedestrian flow record; S602: Based on the pedestrian flow record, record the characteristics and identification information of multiple targets, including appearance characteristics, timestamps, and entry and exit status, provide data support for pedestrian re-identification, and generate pedestrian feature analysis data; S603: Based on the pedestrian feature analysis data, by performing time series analysis on the store's customer flow data, calculating the changing trend of customer flow, evaluating the human resources required for the target store, and generating store customer flow statistics.

[0014] As a further solution of the present invention, the specific formula for calculating the changing trend of passenger flow is: in, Represents at a point in time The trend estimate of is used to predict the passenger flow at the next time point. Represents at a point in time The actual passenger flow, which provides the passenger flow data closest to the current situation. Represents at a point in time Actual passenger flow, provided earlier than passenger flow information, Represents at a point in time The actual passenger flow helps analyze the passenger flow changes at an earlier stage. Represents at a point in time Actual passenger flow, used as a reference for long-term trend analysis, yes The weight coefficient determines the importance of the latest data in the overall trend analysis. yes The weight coefficient adjusts the contribution of the penultimate period data in the prediction model. yes The weight coefficient affects the role of the third-to-last period data in the overall analysis. yes The weight coefficient sets the influence of the earliest data in trend prediction.

[0015] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least: BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 It is a flow chart of a 4S store customer flow statistics algorithm improved based on pedestrian re-identification provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of the refinement of S1 of the present invention; Figure 3 It is a schematic diagram of the refinement of S2 of the present invention; Figure 4 It is a schematic diagram of the refinement of S3 of the present invention; Figure 5 It is a schematic diagram of the refinement of S4 of the present invention; Figure 6 It is a detailed schematic diagram of S5 of the present invention; Figure 7 It is a detailed schematic diagram of S6 of the present invention. DETAILED DESCRIPTION

[0018] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0019] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0020] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same. "of", "corresponding, relevant" and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same.

[0021] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0022] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0023] The embodiment of the present invention provides a 4S store customer flow statistics algorithm based on pedestrian re-identification improvement, such as Figure 1 The flowchart of the 4S store customer flow statistics algorithm based on improved pedestrian re-identification is shown. The processing flow of the method may include the following steps: S1: Based on the image acquisition device, the video data of the 4S store entrance is collected in real time, and the video data is standardized according to the input requirements of video processing to generate the video standardization processing results; S2: Based on the video standardization processing result, the learning speed is configured to match the dynamic changes in the video, the objects in the video data are detected, the size and position of the recognition frame are adjusted to reduce the overlap between multiple frames, and the target detection frame information is obtained; S3: Based on the target detection frame information, by analyzing the position parameters of the identified target, calculating the position change of the target between consecutive frames, identifying the motion path of the target and adding marking information, and generating movement path tracking information; S4: Based on the mobile path tracking information, feature vectors of multiple targets are extracted and compared with feature vectors stored in a database, similarity scores between feature data are calculated, and a re-identification classification result is obtained; S5: Based on the re-identification classification result, an entry detection buffer is set in combination with store layout parameters, the entry and exit behaviors of multiple targets are analyzed in real time, the entry and exit status of multiple targets are updated and recorded, and an area entry detection result is generated; S6: Based on the area entry detection results, count the pedestrians entering and exiting the store entrance, record the characteristics and identification information of multiple pedestrians, evaluate the human resource allocation required for the operation of the target store, and generate store customer flow statistics.

[0024] The video standardization processing result includes image resolution adjustment information, frame rate adjustment record, and timestamp record information; the target detection frame information includes positioning recognition frame size parameters, recognition frame position information, and recognition frame positioning position information; the mobile path tracking information includes the target's movement speed data, movement direction information, and movement trajectory time series; the re-identification classification result includes the target's feature similarity score, a list of targets with confirmed identities, and feature library matching information of targets to be re-identified; the area entry detection result includes the target entry and exit status record, the target's behavior analysis in the buffer zone, and the target's spatial position change record; the store customer flow statistics data includes the counted number of pedestrians, pedestrian features and identification data sets, and customer flow change trend data.

[0025] See also Figure 2 Based on the image acquisition device, the video data of the 4S store entrance is collected in real time. Combined with the input requirements of video processing, the video data is standardized. The specific steps for generating the video standardization processing results are as follows: S101: Based on the image acquisition device, the focal length and viewing angle of the camera are adjusted to optimize the acquisition range, and the video data of the entrance of the 4S store is collected in real time. The time information of the acquisition device is calibrated to optimize the consistency between the video data and the actual time, and the video acquisition result is generated; In sub-step S101, based on the image acquisition device, the device operating parameters are adjusted to improve data quality. The focal length of the camera is adjusted to ensure that vehicles and pedestrians at the entrance of the 4S store are clearly captured. The focal length adjustment depends on the result of calculating the average distance between the camera and the target object. The viewing angle is then adjusted to cover a wider acquisition range. The camera viewing angle adjustment takes into account the entrance width and traffic flow to determine the optimal shooting angle. The time information of the acquisition device is calibrated to ensure the consistency of the video data with the actual time. The calibration process includes synchronizing the internal clock of the device with the network timing service to ensure that each frame of video can be correctly time-stamped. The parameters are executed through the image acquisition control software and fine-tuned according to the image feedback from real-time monitoring to ensure that the acquired video data can reflect the dynamic situation of the 4S store entrance in real time.

[0026] S102: Based on the video acquisition result and according to the input requirements of the video processing, adjusting the resolution of multiple frames of images in the video, optimizing the image clarity, and generating a resolution-adjusted video stream; In sub-step S102, the video acquisition results are processed to match the subsequent video analysis requirements, the resolution of multiple frames of images is adjusted, and the bicubic interpolation method is used to increase the image resolution and retain image details. During the processing, the algorithm reads the image data of each frame, calculates the target resolution that needs to be adjusted, and resamples the pixels of the original image. Each pixel point is interpolated by the surrounding pixel values ​​according to the weight. The generated high-resolution image stream is more suitable for character recognition. The adjusted video stream is stored on the server to ensure that each frame of the image meets the required clarity standard.

[0027] S103: adjusting the video stream based on the resolution, adjusting the frame rate of the video stream, setting the frame rate to match the dynamic scene changes of the video surveillance, and generating a video standardization processing result; In sub-step S103, the frame rate of the video stream after the resolution adjustment is adjusted, and the dynamic scene of the video surveillance is optimized. The frame interpolation technology is used to match the dynamic changes of different scenes. The frame interpolation technology generates intermediate frames by analyzing the action changes between consecutive frames, thereby improving the smoothness of the video stream and monitoring the fast-moving vehicles and people at the entrance. The most appropriate frame rate is calculated and set by the dynamic scene analysis module. The module adjusts the frame rate according to the complexity of the scene and the target movement speed. The frame rate adjustment result directly affects the effect and storage requirements of the video surveillance. The high frame rate ensures that the action is coherent and clear, and the low frame rate reduces the data storage space. The adjusted video standardization processing results are stored in the central monitoring system for security analysis and future review, and the accuracy and efficiency of the processing are guaranteed.

[0028] See also Figure 3 Based on the video standardization processing result, the learning speed is configured to match the dynamic changes in the video, the objects in the video data are detected, the size and position of the recognition frame are adjusted to reduce the overlap between multiple frames, and the steps of obtaining the target detection frame information are specifically as follows: S201: Based on the video normalization processing result, adjusting the learning rate of the target detection model to match the dynamic changes in the video data, optimizing the response speed of the model to various scenes, and generating a learning rate adjustment result; The specific formula for adjusting the learning rate of the target detection model is: in, is the learning rate adjustment function, calculating the current number of iterations The learning rate function combines the cosine annealing algorithm to optimize the dynamic adjustment of the learning rate, ensuring that the learning rate is large at the beginning of training to converge quickly, and gradually decreases near the end of training to refine the adjustment of model parameters. is the current iteration number, indicating the time point in the learning process, and is used to adjust the instantaneous value of the learning rate. is the total number of iterations, defined as the total number of preset iteration cycles in the entire training process, used for normalization ratio to ensure that the learning rate adjustment is reasonably distributed within the training cycle. is the minimum value of the learning rate, ensuring that the learning rate will not drop to an extremely low level that is invalid for model training during the reduction process, thereby maintaining the effectiveness and stability of the algorithm throughout the training process. is pi, a constant introduced in the calculation to ensure that the adjustment of the learning rate has a periodic and smooth transition to adapt to different training stages.

[0029] formula: Detailed explanation of the formula and the process of formula calculation and derivation: The formula is used to dynamically adjust the learning rate to meet the training needs at different stages. The cosine annealing method is used to gradually reduce the learning rate as the number of iterations increases, stabilize at a lower but non-zero learning rate, and refine the model adjustment in the later stages of training. Parameter meaning and setting value: is the current iteration number, assuming that the current number is the 50th iteration, which represents a specific time point in the training process; is the total number of iterations. Assume that the total number of iterations is set to 200, which indicates the length of the training cycle. is the minimum value of the learning rate, assuming it is 0.01, to ensure that the learning rate does not drop to 0 at the end of training, maintaining the effectiveness and stability of the algorithm during training; is the ratio of pi, approximately 3.14159, used in the formula to provide a periodic adjustment factor; Substitute the parameters into the formula for calculation: Calculation results It shows that at the 50th iteration, the learning rate is adjusted to 0.1537, which is a value that is gradually reduced compared to the initial stage, which is in line with the cosine annealing adjustment strategy, ensuring the careful adjustment of the model weights in the middle and late stages of training to avoid overfitting and retaining sufficient learning ability to continue to optimize the model performance. This smooth learning rate reduction strategy helps to achieve a more stable and accurate model training effect.

[0030] S202: Based on the learning rate adjustment result, detecting moving objects in the video frame, recording positions and movement information of multiple targets, and generating target position data; In sub-step S202, moving objects in video frames are detected based on the learning rate adjustment result. The process uses a convolutional neural network in deep learning to identify and record the positions and movement information of multiple targets in the video frame. The YOLO model is used to quickly identify multiple targets and positions in the image in a single viewing of the image. The process includes inputting the adjusted video stream into the model, and the model calculates the probability distribution and position coordinates of the target in each frame through forward propagation. The low-probability detection results are filtered out through non-maximum suppression processing to ensure that the output target position data is highly accurate and reliable. The position and movement information of each detected target is then encoded into a data packet for tracking analysis and subsequent behavior recognition. The generated target position data is used for real-time monitoring and long-term behavior analysis.

[0031] S203: Based on the target position data, the size and position of the recognition frame are adjusted to reduce the overlap between frames, optimize the recognition accuracy and the target tracking capability, and generate target detection frame information; In sub-step S203, based on the target position data, the size and position of the recognition frame are adjusted to reduce the overlap between frames. The bounding box regression technology in image processing is used to automatically adjust the recognition frame according to the actual size of the target to ensure that the frame matches the target size and minimizes overlap. The process includes analyzing the target position data, calculating the spatial distribution around each target, and then dynamically adjusting the size and position of each recognition frame. The IoU calculation standard is used to evaluate the coverage efficiency of the recognition frame before and after adjustment, optimize the recognition frame, improve the accuracy of target recognition, and enhance the model's ability to track targets. The adjusted recognition frame information is used to update the tracking algorithm of the monitoring system, so that the system can effectively identify and track various dynamic targets. The generated target detection frame information is crucial for subsequent security monitoring and behavior analysis.

[0032] See also Figure 4 Based on the target detection frame information, by analyzing the position parameters of the identified target, calculating the position change of the target between consecutive frames, identifying the motion path of the target and adding marking information, the steps of generating the moving path tracking information are specifically as follows: S301: Analyze and identify the position coordinates of multiple targets based on the target detection frame information, compare the coordinate data between consecutive frames, analyze the moving direction and speed of the target in the video, and generate a position change analysis result; In sub-step S301, multiple target position coordinates based on target detection frame information are analyzed, the optical flow method is used to identify and calculate the movement of the target between consecutive video frames, the mode and speed of the object movement in the image sequence are calculated, and the movement speed and direction of the point are estimated by comparing the position change of the same point in two consecutive frames. The process includes extracting the position coordinates of the target in each frame, using the Lucas-Kanade method to estimate the displacement vector of each target between consecutive frames, and calculating the movement speed and direction. The position change data of each target is accurately calculated and compared to ensure the accuracy of the analysis results. The generated position change analysis results describe the movement trajectory of each target in the video, and the information is used for subsequent passenger flow analysis.

[0033] S302: Based on the position change analysis result, the moving paths of multiple targets are tracked in real time, and the moving trajectories and durations of the targets in the video sequence are recorded to generate trajectory recording results; In sub-step S302, the moving paths of multiple targets are tracked in real time based on the position change analysis results, and the moving trajectories and durations in the video sequence are recorded. The process adopts Kalman filter technology, which is used to estimate the state of linear dynamic systems. In video tracking applications, the future position of the target is predicted by setting the initial state and measurement error, and the target position is corrected according to the actual measurement value provided by each frame. The target position is updated in real time. The moving trajectory of each target is continuously updated through an algorithm, and the position and timestamp of each target appearance are recorded to ensure the continuity and accuracy of the trajectory. The generated trajectory recording results provide detailed data support for subsequent analysis, including abnormal behavior detection in security monitoring or customer behavior patterns in retail analysis.

[0034] S303: Based on the trajectory recording result, by associating with the movement trajectory and time stamp of the target, adding identification marks to multiple targets to generate movement path tracking information; In sub-step S303, based on the trajectory recording results, an identification mark is added to each target by associating it with the target's motion trajectory and timestamp. The process uses a target identification algorithm, which assigns a unique identifier to the target based on its frequency of appearance and motion pattern in the video. It includes analyzing the recorded mobile trajectory data, using image recognition technology to identify the characteristics of each target, including color, size or shape, and combining the data with the timestamp to create a continuous tracking mark for each target, enabling the system to continuously track the same target in multiple frames of video, and maintain tracking continuity even when the line of sight is temporarily blocked. The generated mobile path tracking information provides a complete overview of the target movement, providing important data resources for applications such as security monitoring and traffic management.

[0035] See also Figure 5 Based on the mobile path tracking information, the steps of extracting feature vectors of multiple targets and comparing them with feature vectors stored in the database, calculating the similarity scores between feature data, and obtaining the re-identification classification results are as follows: S401: Based on the movement path tracking information, identifying feature vectors of multiple targets according to image information, including size, geometric shape, and color information of the targets, and generating feature vector extraction results; In sub-step S401, feature vectors of multiple targets are extracted based on the mobile path tracking information. The process uses a convolutional neural network in deep learning to identify the size, geometry and color information of the target, including using a pre-trained network model ResNet or VGGNet to process the input image data. The network extracts features at different levels through a multi-layer structure. The features include basic edge, color, shape and texture information. The scattered feature information is integrated into a feature vector of a fixed length through a fully connected layer. The feature vector of each target is calculated and stored for subsequent feature matching. The generated feature vector extraction result includes a detailed descriptive feature of each target, which provides a basis for subsequent matching and analysis.

[0036] S402: Based on the feature vector extraction result, by comparing with the identified targets in the database, calculating the similarity score between the two sets of vectors, and generating a feature matching result; In the above content, use the formula: Calculate the similarity score between the two sets of feature vectors to evaluate the degree of match between the targets; In the formula, Represents the cosine similarity between two vectors, which is used to measure the closeness of the directions of two vectors. For vector and vector No. elements; Detailed explanation of the formula and the process of formula calculation and derivation: Suppose there are two eigenvectors and , , , calculate the similarity score: result This indicates that the two feature vectors have a high similarity, reflecting that the two targets have a high matching probability.

[0037] S403: Based on the feature matching result, by determining the matching status of the target and the known targets in the database, classifying the matched target and updating the identity tag, identifying the re-identification status of the target, and generating a re-identification classification result; In sub-step S403, based on the feature matching results, the target re-identification classification is performed. The process uses a support vector machine classifier to process the matching results and updates the target's identity label. Each target is compared with the known targets in the database based on the similarity score. A threshold is set to determine whether the target matches. If there is a match, the target is classified through the SVM model. The model predicts the classification of new targets based on the labels in the training data. The classification results confirm the re-identification status of each target. The information is used to update the target information in the system database to ensure that the data recorded in the system is kept up to date. The generated re-identification classification results provide accurate target identification and tracking for subsequent monitoring activities and data analysis.

[0038] See also Figure 6 Based on the re-identification classification results, the entry detection buffer is set in combination with the store layout parameters, the entry and exit behaviors of multiple targets are analyzed in real time, and the entry and exit status of multiple targets are updated and recorded. The specific steps of generating the area entry detection results are as follows: S501: Based on the re-identification and classification results, the layout of the target 4S store is analyzed, and boundary parameters of the entrance buffer are configured, including the spatial layout of the store entrance and the direction of pedestrian flow, and a buffer parameter configuration is generated; In sub-step S501, the layout of the 4S store is analyzed based on the re-identification classification results, and the boundary parameters of the entrance buffer are configured. The process uses spatial analysis technology and flow model to quantify the spatial layout of the store entrance and the direction of pedestrian flow. The store floor plan is drawn using CAD software, and the store floor plan is imported through the GIS system. The flow model algorithm, such as flow field simulation, is applied to predict the flow trend of pedestrians under different entrance layouts. The size and shape of the entrance buffer are adjusted according to the simulation results to adapt to the most efficient distribution of pedestrian flow and avoid congestion. The buffer parameter configuration includes the boundary length, width and its position relative to the store entrance. The generated buffer parameter configuration results are recorded and used for subsequent entry and exit management strategies to ensure efficient management and safe monitoring of pedestrian flow.

[0039] S502: Based on the buffer parameter configuration, by real-time analysis of the video data, using the position data to determine and identify the target that crosses the buffer, recording the entry and exit status of the target, and generating a real-time entry and exit status record; In sub-step S502, based on the buffer parameter configuration, the video data is analyzed in real time to identify and record the entry and exit status of the target crossing the buffer zone. The boundary detection algorithm Sobel edge detection is adopted, combined with the object tracking technology multi-target tracking algorithm, and the configured buffer parameters are imported into the video analysis system. The system receives the video stream in real time and uses the boundary detection algorithm to determine the buffer zone boundary. The motion path of each target is tracked by the multi-target tracking algorithm. When the target crosses the set buffer zone boundary, the system automatically marks and records the entry and exit status of the target, including the timestamp of entry or exit, which improves the safety management efficiency of the store and provides real-time data support for subsequent customer flow analysis. The generated real-time entry and exit status records provide detailed dynamic monitoring results for store management.

[0040] S503: According to the real-time entry and exit status record, the target entry and exit data in the database is updated in real time, including the identification code and entry and exit time information of the target, and the area entry detection result is generated; In sub-step S503, the target entry and exit data in the database are updated in real time according to the real-time entry and exit status records, and the real-time database update technology, including triggers and event-driven programming, is used to synchronize the data detected in the video analysis system. The system identifies the identification code of each target and captures the entry and exit time information based on the entry and exit status records obtained from the video analysis. By writing database triggers, once a new entry and exit event is detected, the corresponding entry in the database is automatically updated to ensure the timeliness and accuracy of the database information, support the needs of store management and security monitoring systems, and the generated area entry detection results are fed back to the management personnel in real time through the database, providing instant security alerts and customer flow statistics, thereby enhancing the store's operation and monitoring capabilities.

[0041] See also Figure 7 Based on the area entry detection results, the pedestrians entering and exiting the store entrance are counted, the characteristics and identification information of multiple pedestrians are recorded, the human resource configuration required for the target store operation is evaluated, and the steps of generating store passenger flow statistics are specifically as follows: S601: Based on the area entry detection result, calculate the pedestrian flow at the store entrance, including the number of people entering and leaving the store, and generate a pedestrian flow record; In sub-step S601, the pedestrian flow at the store entrance is calculated based on the regional entry detection results, which involves statistical analysis and data integration. A video analysis system is used to automatically count the number of people entering and leaving the store. A convolutional neural network is used to detect and distinguish human figures to ensure accuracy. In each video frame, the algorithm identifies pedestrians passing through the store entrance and classifies them as "entering" or "leaving". Each time a pedestrian is detected passing through, the system automatically updates the entrance flow counter and records the timestamp to achieve real-time flow monitoring. Historical data is stored in log files for long-term trend analysis. The generated pedestrian flow records illustrate the number of people entering and leaving the store every day, providing data support for store operation decisions.

[0042] S602: Based on the pedestrian flow record, record the characteristics and identification information of multiple targets, including appearance characteristics, timestamps, and entry and exit status, provide data support for pedestrian re-identification, and generate pedestrian feature analysis data; In sub-step S602, based on the pedestrian flow records, the features and identification information of multiple targets are recorded and analyzed. The system uses facial recognition technology and feature extraction algorithms to identify the appearance features of each pedestrian passing through the store entrance. The deep learning model FaceNet is used to extract high-dimensional feature vectors from the pedestrian images captured by the video. For each identified individual, its appearance features, timestamp, and entry and exit status are recorded, and the information is stored in the database to support real-time security monitoring and provide data resources for pedestrian re-identification tasks. The generated pedestrian feature analysis data provides a reference for subsequent customer behavior analysis and security management.

[0043] S603: Based on the pedestrian feature analysis data, by performing time series analysis on the store's customer flow data, calculating the change trend of customer flow, evaluating the manpower allocation required for the target store, and generating store customer flow statistics; The specific formula for calculating the changing trend of passenger flow is: in, Represents at a point in time The trend estimate of is used to predict the passenger flow at the next time point. Represents at a point in time The actual passenger flow, which provides the passenger flow data closest to the current situation. Represents at a point in time Actual passenger flow, provided earlier than passenger flow information, Represents at a point in time The actual passenger flow helps analyze the passenger flow changes at an earlier stage. Represents at a point in time Actual passenger flow, used as a reference for long-term trend analysis, yes The weight coefficient determines the importance of the latest data in the overall trend analysis. yes The weight coefficient adjusts the contribution of the penultimate period data in the prediction model. yes The weight coefficient affects the role of the third-to-last period data in the overall analysis. yes The weight coefficient sets the influence of the earliest data in trend prediction.

[0044] formula: Detailed explanation of the formula and the process of formula calculation and derivation: The formula is used to calculate the predicted customer flow trend value of the store at the current time point; Parameter meaning and setting value: Indicates time point The actual passenger flow data is assumed to be 120, 115, 110, and 105 passengers; Represents the weight coefficient of the data at each time point, assumed to be 0.4, 0.3, 0.2, and 0.1, reflecting that the data closer to the current time point has a greater impact on trend prediction.

[0045] Substitute the parameters into the formula for calculation: The results showed that 115 people The predicted customer flow is 115 people. The value is calculated based on the actual customer flow data of the past four times and their respective weights. It is used to predict the future customer flow trend and help store management to rationally allocate manpower and resources.

[0046] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. An improved 4S store customer flow statistics algorithm based on pedestrian re-identification, characterized in that: The algorithm includes: Based on the image acquisition equipment, the video data of the 4S store entrance is collected in real time, and the video data is standardized according to the input requirements of video processing to generate video standardization processing results; Based on the video standardization processing result, the learning speed is configured to match the dynamic changes in the video, the objects in the video data are detected, the size and position of the recognition frame are adjusted to reduce the overlap between multiple frames, and the target detection frame information is obtained; Based on the target detection frame information, by analyzing the position parameters of the identified target, calculating the position change of the target between consecutive frames, identifying the motion path of the target and adding marking information, and generating movement path tracking information; Based on the mobile path tracking information, feature vectors of multiple targets are extracted and compared with feature vectors stored in a database, similarity scores between feature data are calculated, and a re-identification classification result is obtained; Based on the re-identification classification results, the entry detection buffer is set in combination with the store layout parameters, the entry and exit behaviors of multiple targets are analyzed in real time, the entry and exit status of multiple targets are updated and recorded, and the area entry detection results are generated; Based on the area entry detection results, the pedestrians entering and exiting the store entrance are counted, the characteristics and identification information of multiple pedestrians are recorded, the human resource allocation required for the operation of the target store is evaluated, and store customer flow statistics are generated.

2. The 4S store customer flow statistics algorithm based on improved pedestrian re-identification according to claim 1 is characterized in that: The video standardization processing result includes image resolution adjustment information, frame rate adjustment record, and timestamp record information; the target detection frame information includes positioning recognition frame size parameters, recognition frame position information, and recognition frame positioning position information; the mobile path tracking information includes the target's movement speed data, movement direction information, and movement trajectory time series; the re-identification classification result includes the target's feature similarity score, a list of targets with confirmed identities, and feature library matching information of targets to be re-identified; the area entry detection result includes the target entry and exit status record, the target's behavior analysis in the buffer zone, and the target's spatial position change record; the store customer flow statistics data includes the counted number of pedestrians, pedestrian features and identification data sets, and customer flow change trend data.

3. The 4S store customer flow statistics algorithm based on improved pedestrian re-identification according to claim 1 is characterized in that: Based on the image acquisition device, the video data of the 4S store entrance is collected in real time. Combined with the input requirements of video processing, the video data is standardized. The specific steps to generate the video standardization processing results are as follows: Based on the image acquisition device, by adjusting the focal length and viewing angle of the camera, optimizing the acquisition range, real-time acquisition of video data of the 4S store entrance, and calibrating the time information of the acquisition device, optimizing the consistency between the video data and the actual time, and generating video acquisition results; Based on the video acquisition result and according to the input requirements of the video processing, the resolution of multiple frames of images in the video is adjusted to optimize the image clarity and generate a resolution-adjusted video stream; The video stream is adjusted based on the resolution, the frame rate of the video stream is adjusted, the frame rate is set to match the dynamic scene changes of the video surveillance, and a video standardization processing result is generated.

4. The 4S store customer flow statistics algorithm based on improved pedestrian re-identification according to claim 1 is characterized in that: Based on the video normalization processing result, the steps of detecting objects in the video data by configuring the learning speed to match the dynamic changes in the video, adjusting the size and position of the recognition frame to reduce the overlap between multiple frames, and obtaining the target detection frame information are as follows: Based on the video normalization processing result, adjusting the learning rate of the target detection model to match the dynamic changes in the video data, optimizing the response speed of the model to various scenes, and generating a learning rate adjustment result; Based on the learning rate adjustment result, detecting moving objects in the video frame, recording the positions and movement information of multiple targets, and generating target position data; Based on the target position data, the overlap between frames is reduced by adjusting the size and position of the recognition frame, thereby optimizing the recognition accuracy and the target tracking capability, and generating target detection frame information.

5. The 4S store customer flow statistics algorithm based on improved pedestrian re-identification according to claim 4 is characterized in that: The specific formula for adjusting the learning rate of the target detection model is: in, is the learning rate adjustment function, calculating the current number of iterations The learning rate function combines the cosine annealing algorithm to optimize the dynamic adjustment of the learning rate, ensuring that the learning rate is large at the beginning of training to converge quickly, and gradually decreases near the end of training to refine the adjustment of model parameters. is the current iteration number, indicating the time point in the learning process, and is used to adjust the instantaneous value of the learning rate. is the total number of iterations, defined as the total number of preset iteration cycles in the entire training process, used for normalization ratio to ensure that the learning rate adjustment is reasonably distributed within the training cycle. is the minimum value of the learning rate, ensuring that the learning rate will not drop to an extremely low level that is invalid for model training during the reduction process, thereby maintaining the effectiveness and stability of the algorithm throughout the training process. is pi, a constant introduced in the calculation to ensure that the adjustment of the learning rate has a periodic and smooth transition to adapt to different training stages.

6. The 4S store customer flow statistics algorithm based on improved pedestrian re-identification according to claim 1 is characterized in that: Based on the target detection frame information, by analyzing the position parameters of the identified target, calculating the position change of the target between consecutive frames, identifying the motion path of the target and adding marking information, the steps of generating the moving path tracking information are specifically as follows: Based on the target detection frame information, analyze and identify the position coordinates of multiple targets, compare the coordinate data between consecutive frames, analyze the moving direction and speed of the target in the video, and generate a position change analysis result; Based on the position change analysis results, the moving paths of multiple targets are tracked in real time, and the moving trajectories and durations of the targets in the video sequence are recorded to generate trajectory recording results; Based on the trajectory recording results, identification marks are added to multiple targets by associating with the motion trajectories and timestamps of the targets to generate movement path tracking information.

7. The 4S store customer flow statistics algorithm based on improved pedestrian re-identification according to claim 1 is characterized in that: Based on the mobile path tracking information, the steps of extracting feature vectors of multiple targets and comparing them with feature vectors stored in the database, calculating similarity scores between feature data, and obtaining re-identification classification results are as follows: Based on the movement path tracking information, identifying feature vectors of multiple targets according to image information, including size, geometry, and color information of the targets, and generating feature vector extraction results; Based on the feature vector extraction result, by comparing with the identified targets in the database, the similarity score between the two sets of vectors is calculated to generate a feature matching result; Based on the feature matching result, by judging the matching status of the target and the known targets in the database, the matched target is classified and the identity tag is updated, the re-identification status of the target is identified, and the re-identification classification result is generated.

8. The improved 4S store customer flow statistics algorithm based on pedestrian re-identification according to claim 1 is characterized in that: Based on the re-identification classification results, the steps of setting the entry detection buffer in combination with the store layout parameters, analyzing the entry and exit behaviors of multiple targets in real time, updating and recording the entry and exit status of multiple targets, and generating the area entry detection results are as follows: Based on the re-identification classification results, the layout of the target 4S store is analyzed, and boundary parameters of the entrance buffer are configured, including the spatial layout of the store entrance and the direction of pedestrian flow, and a buffer parameter configuration is generated; Based on the buffer parameter configuration, by real-time analysis of the video data, using the position data to determine and identify the target that crosses the buffer, recording the entry and exit status of the target, and generating a real-time entry and exit status record; According to the real-time entry and exit status record, the target entry and exit data in the database is updated in real time, including the identification code and entry and exit time information for identifying the target, and the area entry detection result is generated.

9. The 4S store customer flow statistics algorithm based on improved pedestrian re-identification according to claim 1 is characterized in that: Based on the area entry detection results, the pedestrians entering and exiting the store entrance are counted, the characteristics and identification information of multiple pedestrians are recorded, the human resource configuration required for the target store operation is evaluated, and the steps of generating store customer flow statistics are specifically as follows: Based on the area entry detection results, the pedestrian flow at the store entrance is calculated, including the number of people entering and leaving the store, and a pedestrian flow record is generated; Based on the pedestrian flow record, the characteristics and identification information of multiple targets are recorded, including appearance characteristics, timestamps, and entry and exit status, to provide data support for pedestrian re-identification and generate pedestrian feature analysis data; Based on the pedestrian feature analysis data, by performing time series analysis on the store's customer flow data, the changing trend of customer flow is calculated, the human resource allocation required for the target store is evaluated, and store customer flow statistics are generated.

10. The 4S store customer flow statistics algorithm based on improved pedestrian re-identification according to claim 9 is characterized in that: The specific formula for calculating the changing trend of passenger flow is: in, Represents at a point in time The trend estimate of is used to predict the passenger flow at the next time point. Represents at a point in time The actual passenger flow, which provides the passenger flow data closest to the current situation. Represents at a point in time Actual passenger flow, provided earlier than passenger flow information, Represents at a point in time The actual passenger flow helps analyze the passenger flow changes at an earlier stage. Represents at a point in time Actual passenger flow, used as a reference for long-term trend analysis, yes The weight coefficient determines the importance of the latest data in the overall trend analysis. yes The weight coefficient adjusts the contribution of the penultimate period data in the prediction model. yes The weight coefficient affects the role of the third-to-last period data in the overall analysis. yes The weight coefficient sets the influence of the earliest data in trend prediction.

Images