Crowd behavior intelligent perception analysis method and device based on computer vision, equipment, storage medium and product

Abstract

The invention discloses a crowd behavior intelligent perception analysis method and device based on computer vision, equipment, a storage medium and a product, and relates to the technical field of data analysis, and the method comprises the steps: obtaining a target video file corresponding to a target crowd, and recognizing the crowd flow distribution data of the target crowd from the target video file; tracking and identifying the target crowd according to the crowd flow distribution data to obtain crowd behavior data, the crowd behavior data including crowd distance, crowd residence time, crowd density and crowd trajectory; generating a conventional thermodynamic analysis chart corresponding to the target crowd based on the crowd behavior data; and performing inverse perspective transformation on the conventional thermodynamic analysis chart to obtain a forward overlook thermodynamic analysis chart, and performing crowd behavior intelligent perception analysis on the target crowd based on the forward overlook thermodynamic analysis chart. According to the method, crowd behavior intelligent perception analysis is performed on the target crowd based on the forward overlooking thermodynamic analysis chart, so that the user can intuitively analyze various crowd behaviors from the forward overlooking view angle.

Description

Technical Field

[0001] The present application relates to the field of data analysis technology, and in particular to a method, device, equipment, storage medium and product for intelligent perception analysis of crowd behavior based on computer vision. Background Art

[0002] Currently, crowd behavior perception based on computer vision is an important means for judging regional vitality in the industry. Existing methods often use cameras to capture crowd behavior and then identify and analyze it. However, due to the limitations of the camera's shooting angle, crowd counting or crowd flow diagram generation can only be performed within the perspective image within the camera's viewfinder, which makes it impossible for users to intuitively perform intelligent crowd behavior perception analysis based on computer vision on the target population from a bird's-eye view. Summary of the invention

[0003] The main purpose of this application is to provide a method, device, equipment, storage medium and product for intelligent perception analysis of crowd behavior based on computer vision, aiming to solve the technical problem that the existing technology cannot intuitively perform intelligent perception analysis of crowd behavior based on computer vision on the target population from a bird's-eye view.

[0004] To achieve the above objectives, the present application provides a method for intelligent perception analysis of crowd behavior based on computer vision, the method comprising the following steps:

[0005] Obtaining a target video file corresponding to a target population, and identifying crowd distribution data of the target population from the target video file;

[0006] Tracking and identifying the target crowd according to the crowd flow distribution data to obtain crowd behavior data, wherein the crowd behavior data includes crowd distance, crowd residence time, crowd density and crowd trajectory;

[0007] Generate a conventional thermodynamic analysis diagram corresponding to the target population based on the population behavior data;

[0008] The conventional thermodynamic analysis diagram is subjected to an inverse perspective transformation to obtain an overhead view thermodynamic analysis diagram, and crowd behavior intelligent perception analysis is performed on the target population based on the overhead view thermodynamic analysis diagram.

[0009] In one embodiment, the step of identifying the crowd distribution data of the target group from the target video file includes:

[0010] The target video file is scanned and identified frame by frame by using the YOLOX algorithm and the ByteTrack algorithm to obtain a number of video frames;

[0011] The plurality of video frames are color-labeled and number-labeled to obtain crowd flow distribution data corresponding to the target population.

[0012] In one embodiment, the step of tracking and identifying the target group of people according to the crowd distribution data to obtain crowd behavior data includes:

[0013] Performing regional screening on the crowd flow distribution data based on a preset analysis target to obtain screened crowd flow distribution data;

[0014] The target population is tracked and identified according to the crowd information and crowd behavior corresponding to the target population in the filtered crowd flow distribution data to obtain crowd behavior data.

[0015] In one embodiment, the conventional thermodynamic analysis graph includes a crowd distance distribution thermodynamic graph, a crowd occupancy thermodynamic analysis graph, a crowd density distribution thermodynamic graph, and a crowd trajectory distribution thermodynamic graph. The step of generating the conventional thermodynamic analysis graph corresponding to the target crowd based on the crowd behavior data includes:

[0016] Generate a crowd distance distribution heat map corresponding to the target crowd based on the crowd distance, wherein the crowd distance is obtained by tracking the pedestrian landing data identified frame by frame;

[0017] A crowd occupancy rate heat analysis diagram corresponding to the target crowd is generated based on the crowd residence time, where the crowd residence time is the time that pedestrians tracked and identified at the same point in the target video file stay.

[0018] In one embodiment, the step of generating a conventional thermodynamic analysis diagram corresponding to the target population based on the population behavior data further includes:

[0019] Generate a crowd density distribution heat map corresponding to the target crowd based on the crowd density, where the crowd density is the number of pedestrians passing by the same point in the target video file;

[0020] A crowd trajectory distribution heat map corresponding to the target crowd is generated based on the crowd trajectory, where the crowd trajectory is the movement trajectory of pedestrians in the target video file.

[0021] In one embodiment, the step of performing an inverse perspective transformation on the conventional thermodynamic analysis diagram to obtain a top-view thermodynamic analysis diagram, and performing crowd behavior intelligent perception analysis on the target population based on the top-view thermodynamic analysis diagram, comprises:

[0022] Taking any rectangular area in the target video file as a perspective reference, and performing an inverse perspective transformation on the conventional thermal analysis diagram based on the perspective reference to obtain a top-down thermal analysis diagram;

[0023] Based on the top-view thermal analysis diagram, intelligent crowd behavior perception analysis is performed on the target population to obtain the crowd flow vitality distribution in the area where the target population is located.

[0024] In addition, to achieve the above purpose, the present application also proposes a crowd behavior intelligent perception analysis device based on computer vision, and the crowd behavior intelligent perception analysis device based on computer vision includes:

[0025] A data acquisition module, used to acquire a target video file corresponding to a target group of people, and identify crowd distribution data of the target group of people from the target video file;

[0026] A tracking and identification module is used to track and identify the target crowd according to the crowd flow distribution data to obtain crowd behavior data, wherein the crowd behavior data includes crowd distance, crowd residence time, crowd density and crowd trajectory;

[0027] An image generation module, used for generating a conventional thermodynamic analysis diagram corresponding to the target population based on the population behavior data;

[0028] The transformation analysis module is used to perform an inverse perspective transformation on the conventional thermal analysis diagram to obtain a top-view thermal analysis diagram, and perform intelligent perception analysis of crowd behavior on the target population based on the top-view thermal analysis diagram.

[0029] In addition, to achieve the above-mentioned purpose, the present application also proposes a computer vision-based crowd behavior intelligent perception analysis device, the device comprising: a memory, a processor, and a computer vision-based crowd behavior intelligent perception analysis program stored in the memory and executable on the processor, the computer vision-based crowd behavior intelligent perception analysis program being configured to implement the steps of the computer vision-based crowd behavior intelligent perception analysis method as described above.

[0030] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and on which is stored a computer vision-based intelligent perception analysis program for crowd behavior. When the computer vision-based intelligent perception analysis program for crowd behavior is executed by a processor, the steps of the computer vision-based intelligent perception analysis method for crowd behavior are implemented as described above.

[0031] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer program product, which includes a crowd behavior intelligent perception analysis program based on computer vision, and when the crowd behavior intelligent perception analysis program based on computer vision is executed by a processor, it implements the steps of the crowd behavior intelligent perception analysis method based on computer vision as described above.

[0032] This application obtains the target video file corresponding to the target population, and identifies the crowd distribution data of the target population from the target video file; tracks and identifies the target population according to the crowd distribution data to obtain crowd behavior data, and the crowd behavior data includes crowd distance, crowd residence time, crowd density and crowd trajectory; generates a conventional thermal analysis diagram corresponding to the target population based on the crowd behavior data; performs an inverse perspective transformation on the conventional thermal analysis diagram to obtain an overhead view thermal analysis diagram, and performs crowd behavior intelligent perception analysis on the target population based on the overhead view thermal analysis diagram. Compared with the traditional crowd behavior analysis method, since the above method of this application performs an inverse perspective transformation on the conventional thermal analysis diagram of the target population to obtain an overhead view thermal analysis diagram, it is possible to perform crowd behavior intelligent perception analysis on the target population based on the overhead view thermal analysis diagram, so that the user can intuitively analyze various types of crowd behaviors from an overhead view perspective, and can use the same video file to perform spatial or temporal division on the same space to comprehensively study the relationship between crowd activities and space in the space. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A schematic diagram of the structure of a crowd behavior intelligent perception and analysis device based on computer vision in the hardware operating environment involved in the embodiment of the present application;

[0034] Figure 2 This is a flowchart of the first embodiment of the crowd behavior intelligent perception analysis method based on computer vision of this application;

[0035] Figure 3 This is a schematic diagram of the BodyCAM software interface in the computer vision-based crowd behavior intelligent perception analysis method of this application;

[0036] Figure 4 This is a flow chart of the second embodiment of the method for intelligent perception and analysis of crowd behavior based on computer vision of this application;

[0037] Figure 5 This is a flowchart of the third embodiment of the method for intelligent perception and analysis of crowd behavior based on computer vision of this application;

[0038] Figure 6 This is a structural block diagram of the first embodiment of the computer vision-based intelligent perception and analysis device for crowd behavior of this application.

[0039] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0040] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0041] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of a computer vision-based crowd behavior intelligent perception and analysis device in the hardware operating environment involved in the embodiment of the present application.

[0042] like Figure 1 As shown, the crowd behavior intelligent perception analysis device based on computer vision may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (Wireless-Fidelity, Wi-Fi) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM), or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk storage. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0043] Those skilled in the art will understand that Figure 1 The structure shown in the figure does not constitute a limitation on the crowd behavior intelligent perception and analysis device based on computer vision, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0044] like Figure 1 As shown, the memory 1005 as a storage medium may include an operating system, a network communication module, a user interface module, and a crowd behavior intelligent perception analysis program based on computer vision.

[0045] exist Figure 1In the computer vision-based intelligent perception and analysis device for crowd behavior shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the computer vision-based intelligent perception and analysis device for crowd behavior of the present application can be set in the computer vision-based intelligent perception and analysis device for crowd behavior, and the computer vision-based intelligent perception and analysis device for crowd behavior calls the computer vision-based intelligent perception and analysis program for crowd behavior stored in the memory 1005 through the processor 1001, and executes the computer vision-based intelligent perception and analysis method for crowd behavior provided in the embodiment of the present application.

[0046] The present application embodiment provides a method for intelligent perception and analysis of crowd behavior based on computer vision, referring to Figure 2 , Figure 2 This is a flow chart of the first embodiment of the computer vision-based intelligent perception and analysis method of crowd behavior in this application.

[0047] In this embodiment, the crowd behavior intelligent perception analysis method based on computer vision includes the following steps:

[0048] Step S10: Acquire a target video file corresponding to a target group, and identify crowd distribution data of the target group from the target video file.

[0049] It should be noted that the execution subject of the method of this embodiment can be a terminal device with image generation, data processing and program running functions, such as a smart terminal, a computer, etc., or an electronic device with the same or similar functions, such as the above-mentioned crowd behavior intelligent perception analysis device based on computer vision. The following takes the crowd behavior intelligent perception analysis device based on computer vision (hereinafter referred to as the analysis device) as an example to illustrate this embodiment and the following embodiments.

[0050] It is understandable that the above analysis device is installed with BodyCAM software, which is a video tracking tool based on video clips that can track pedestrian trajectories, identify geometric boundaries, and analyze pedestrian behavior patterns and spatial relationships. Figure 3 , Figure 3 This is a schematic diagram of the BodyCAM software interface in the computer vision-based intelligent perception and analysis method of crowd behavior in this application.

[0051] It should be understood that the target video file may be in MP4 format, or in AVI format, MKV format, FLV format, etc., and this embodiment does not impose any limitation on this.

[0052] In a specific implementation, after the target video file corresponding to the target population is determined, the target video file can be input into the above-mentioned BodyCAM software for identification, so as to obtain the crowd distribution data of the target population.

[0053] Step S20: Track and identify the target crowd according to the crowd distribution data to obtain crowd behavior data, which includes crowd distance, crowd residence time, crowd density and crowd trajectory.

[0054] It should be understood that the pedestrian numbers corresponding to the target crowd in the target video file can be determined based on the crowd distribution data, and then each pedestrian number can be tracked and identified at the same time to obtain the above-mentioned crowd behavior data including crowd distance, crowd residence time, crowd density and crowd trajectory.

[0055] Step S30: Generate a conventional thermodynamic analysis diagram corresponding to the target population based on the population behavior data.

[0056] It should be noted that the above conventional thermodynamic analysis diagrams may include a crowd distance distribution thermodynamic diagram corresponding to the target population, a crowd occupancy thermodynamic analysis diagram, a crowd density distribution thermodynamic diagram, a crowd trajectory distribution thermodynamic diagram, etc., and this embodiment does not limit this.

[0057] Step S40: performing an inverse perspective transformation on the conventional thermodynamic analysis diagram to obtain an overhead view thermodynamic analysis diagram, and performing crowd behavior intelligent perception analysis on the target population based on the overhead view thermodynamic analysis diagram.

[0058] It should be noted that the above-mentioned top-down thermal analysis diagram refers to a thermal diagram at a top-down angle.

[0059] In a specific implementation, the conventional thermal analysis diagram can be inversely transformed based on the Inverse Perspective Mapping algorithm to obtain the above-mentioned top-view thermal analysis diagram. Inverse Perspective Mapping (IPM for short) can convert an image with perspective effect into a more intuitive and easy-to-analyze image form, that is, an image after eliminating the perspective effect, which is usually called a bird's eye view or a top view.

[0060] This embodiment obtains the target video file corresponding to the target population, and identifies the crowd distribution data of the target population from the target video file; tracks and identifies the target population according to the crowd distribution data to obtain crowd behavior data, and the crowd behavior data includes crowd distance, crowd residence time, crowd density and crowd trajectory; generates a conventional thermal analysis diagram corresponding to the target population based on the crowd behavior data; performs an inverse perspective transformation on the conventional thermal analysis diagram to obtain an overhead view thermal analysis diagram, and performs crowd behavior intelligent perception analysis on the target population based on the overhead view thermal analysis diagram. Compared with the traditional crowd behavior analysis method, since the above method of this embodiment performs an inverse perspective transformation on the conventional thermal analysis diagram of the target population to obtain an overhead view thermal analysis diagram, it is possible to perform crowd behavior intelligent perception analysis on the target population based on the overhead view thermal analysis diagram, so that the user can intuitively analyze various types of crowd behaviors from an overhead view perspective, and can use the same video file to perform spatial division or temporal division on the same space to comprehensively study the relationship between crowd activities and space in the space.

[0061] refer to Figure 4 , Figure 4 This is a flow chart of the second embodiment of the computer vision-based intelligent perception and analysis method of crowd behavior in this application.

[0062] In a feasible implementation manner, the step S10 may include:

[0063] Step S101: Scan and identify the target video file frame by frame using the YOLOX algorithm and the ByteTrack algorithm to obtain a number of video frames.

[0064] It should be noted that the above-mentioned YOLOX algorithm is a lightweight target detection algorithm. YOLOX adopts the design idea of ​​decoupling head, and entrusts category prediction and position prediction to two different networks respectively; the above-mentioned ByteTrack algorithm is an algorithm for multi-target tracking. ByteTrack processes the association between the detection frame and the tracking trajectory through two matching processes. First, the high-score detection frame is matched, and then the unmatched high-score detection frame and low-score detection frame are matched twice to restore the occluded or blurred targets.

[0065] In the specific implementation, the OpenCV library can be used to load the target video file frame by frame, and then the YOLOX algorithm can be used to scan and detect the loaded target video file frame by frame, and the ByteTrack algorithm can be used to track and identify the detection results, and finally the above-mentioned video frames can be obtained.

[0066] Step S102: color-labeling and number-labeling the plurality of video frames to obtain crowd flow distribution data corresponding to the target population.

[0067] In the specific implementation, you can use a green rectangle for color marking, and use the number in the yellow box in the upper left corner of the green box to indicate the pedestrian number for number marking. By comparing the crowd recognition results of the previous and next frames, you can identify and track the behavior of people with specific numbers. In addition, users can also click the "PAUSE" option in the BodyCAM software interface to pause the video and view the frame-by-frame tracking analysis results.

[0068] In a feasible implementation manner, the step S20 may include:

[0069] Step S201: performing regional screening on the crowd flow distribution data based on a preset analysis target to obtain screened crowd flow distribution data.

[0070] Step S202: Tracking and identifying the target population according to the crowd information and crowd behavior corresponding to the target population in the filtered crowd flow distribution data to obtain crowd behavior data.

[0071] In the specific implementation, for children and adults, the length-to-width ratio of the corresponding green rectangle is different. The length-to-width ratio of adults is generally much larger than that of children. Users can control the length-to-width ratio of the tracking subject through the "Track by Ratio" option in the BodyCAM software interface to achieve the purpose of tracking children and adults separately. In addition, users can also define the recognition area in the target group video by clicking the "AREA" option in the BodyCAM software interface. The software can track and identify the crowd behavior data in the selected area.

[0072] This embodiment uses the YOLOX algorithm and the ByteTrack algorithm to scan and identify the target video file frame by frame to obtain several video frames; the several video frames are color-labeled and numbered to obtain the crowd distribution data corresponding to the target population; the crowd distribution data is regionally screened based on the preset analysis target to obtain the screened crowd distribution data; the target population is tracked and identified according to the crowd information and crowd behavior corresponding to the target population in the screened crowd distribution data to obtain crowd behavior data. Compared with the traditional crowd behavior analysis method, the above method of this embodiment uses the YOLOX algorithm and the ByteTrack algorithm to scan and identify the target video file frame by frame, thereby improving the recognition accuracy; at the same time, the crowd distribution data is screened based on the preset analysis target, so that the analysis results of the crowd in this embodiment are more targeted.

[0073] refer to Figure 5 , Figure 5This is a flow chart of the third embodiment of the computer vision-based intelligent perception and analysis method of crowd behavior in this application.

[0074] In a feasible implementation, the conventional thermodynamic analysis diagram includes a crowd distance distribution thermodynamic diagram, a crowd occupancy thermodynamic analysis diagram, a crowd density distribution thermodynamic diagram, and a crowd trajectory distribution thermodynamic diagram. The step S30 may include:

[0075] Step S301: generating a crowd distance distribution heat map corresponding to the target crowd based on the crowd distance, wherein the crowd distance is obtained by tracking and identifying pedestrian footsteps frame by frame.

[0076] Step S302: Generate a crowd occupancy rate heat analysis diagram corresponding to the target crowd based on the crowd residence time, where the crowd residence time is the time that the pedestrians tracked and identified at the same point in the target video file stay.

[0077] Step S303: generating a crowd density distribution heat map corresponding to the target crowd based on the crowd density, wherein the crowd density is the number of pedestrians passing by the same point in the target video file.

[0078] Step S304: generating a crowd trajectory distribution heat map corresponding to the target crowd based on the crowd trajectory, wherein the crowd trajectory is the movement trajectory of pedestrians in the target video file.

[0079] In the specific implementation, you can click the "Heatmap:Distance" option in the BodyCAM software interface to generate a crowd distance distribution heat map. The crowd distance distribution heat map is generated by superimposing the crowd flow data tracked and identified frame by frame. The crowd distance distribution heat map represents the distance between people. The smaller the distance between the crowd flow data obtained by tracking and identifying, the denser the crowd, the redder the color, and vice versa. Click the "Heatmap:Duration" option in the BodyCAM software interface to generate a crowd occupancy heat analysis diagram. The crowd occupancy heat analysis diagram is generated by calculating the stay time of the crowd tracked and identified at the same point. The longer the crowd stays, the redder the color, and vice versa. Click the "Heatmap:Duration" option in the BodyCAM software interface to generate a crowd occupancy heat analysis diagram. Click the "Trajactories" option in the BodyCAM software interface to output a heat map of crowd density distribution. The more people pass by the same point, the redder the color of the point, and vice versa. Click the "Trajactories" option in the BodyCAM software interface to output a heat map of crowd trajectory distribution. Different colors in the map represent the speed differences when identifying pedestrians in motion. The faster the speed, the brighter the color. If you click the "AREA" option in the BodyCAM software interface to divide the area, you can track and identify the people in the area frame by frame to perform spatiotemporal counting (spatiotemporal counting refers to the statistical data of the number of people and occupancy time in a designated area over a period of time, which can be visualized as a table to reflect the usage of different areas in different time periods), and output the crowd spatiotemporal counting results as a table.

[0080] In a feasible implementation manner, the step S40 may include:

[0081] Step S401: taking any rectangular area in the target video file as a perspective reference, and performing an inverse perspective transformation on the conventional thermal analysis diagram based on the perspective reference to obtain a top-down thermal analysis diagram.

[0082] It should be noted that the above-mentioned perspective reference object refers to a rectangular reference object on the real-world working plane. According to the perspective change of this rectangle, the visual data under the perspective condition can be reversely converted into an image under the top view angle without perspective.

[0083] Step S402: Based on the top-view thermal analysis diagram, intelligent crowd behavior perception analysis is performed on the target population to obtain the crowd flow vitality distribution in the area where the target population is located.

[0084] In the specific implementation, the analysis of crowd behavior mainly relies on the frame-by-frame analysis of video files. Therefore, as long as the number of frames reaches a certain level, crowd behavior analysis can be performed (the heat map is the result of the analysis of the target population, and the heat map is the result image of the superimposed output of the target population behavior analysis calculation within a period of time). That is, an analysis file can be output at every fixed frame interval, or analysis files of different time periods under the same video file can be output. The user can enter and check the "Output by Frame Number" option in the operation interface to set the interval frame number of the exported file. For example, if you enter 100, you can export a file every 100 frames, and you can click the "Calculation" option to select the output file path to automatically output the file.

[0085] This embodiment generates a crowd distance distribution heat map corresponding to the target crowd based on the crowd distance, and the crowd distance is obtained by tracking and identifying the pedestrian landing data frame by frame; generates a crowd occupancy rate heat analysis map corresponding to the target crowd based on the crowd residence time, and the crowd residence time is the time that the pedestrians tracked and identified at the same point in the target video file stay; generates a crowd density distribution heat map corresponding to the target crowd based on the crowd density, and the crowd density is the number of pedestrians passing by the same point in the target video file; generates a crowd trajectory distribution heat map corresponding to the target crowd based on the crowd trajectory, and the crowd trajectory is the movement trajectory of the pedestrian in the target video file; takes any rectangular area in the target video file as a perspective reference, and performs an inverse perspective transformation on the conventional heat analysis map based on the perspective reference to obtain a top-down view heat analysis map; performs crowd behavior intelligent perception analysis on the target crowd based on the top-down view heat analysis map to obtain the distribution of pedestrian flow vitality in the area where the target crowd is located. Compared with traditional crowd behavior analysis methods, the above method of this embodiment performs intelligent crowd behavior perception analysis based on computer vision on the target crowd based on the crowd distance distribution heat map, crowd occupancy rate heat analysis map, crowd density distribution heat map and crowd trajectory distribution heat map under a direct bird's-eye view, so that users can intuitively analyze various types of crowd behaviors from a direct bird's-eye view perspective.

[0086] In addition, an embodiment of the present application also proposes a storage medium, on which is stored a computer vision-based intelligent perception and analysis program for crowd behavior. When the computer vision-based intelligent perception and analysis program for crowd behavior is executed by a processor, the steps of the computer vision-based intelligent perception and analysis method for crowd behavior as described above are implemented.

[0087] Reference Figure 6 , Figure 6 This is a structural block diagram of the first embodiment of the computer vision-based intelligent perception and analysis device for crowd behavior of this application.

[0088] like Figure 6 As shown, the crowd behavior intelligent perception and analysis device based on computer vision proposed in the embodiment of the present application includes:

[0089] The data acquisition module 601 is used to acquire a target video file corresponding to a target group, and identify crowd distribution data of the target group from the target video file;

[0090] A tracking and identification module 602 is used to track and identify the target crowd according to the crowd flow distribution data to obtain crowd behavior data, wherein the crowd behavior data includes crowd distance, crowd residence time, crowd density and crowd trajectory;

[0091] An image generation module 603 is used to generate a conventional thermodynamic analysis diagram corresponding to the target population based on the population behavior data;

[0092] The transformation analysis module 604 is used to perform an inverse perspective transformation on the conventional thermodynamic analysis diagram to obtain an overhead view thermodynamic analysis diagram, and perform crowd behavior intelligent perception analysis on the target population based on the overhead view thermodynamic analysis diagram.

[0093] This embodiment obtains the target video file corresponding to the target population, and identifies the crowd distribution data of the target population from the target video file; tracks and identifies the target population according to the crowd distribution data to obtain crowd behavior data, and the crowd behavior data includes crowd distance, crowd residence time, crowd density and crowd trajectory; generates a conventional thermal analysis diagram corresponding to the target population based on the crowd behavior data; performs an inverse perspective transformation on the conventional thermal analysis diagram to obtain an overhead view thermal analysis diagram, and performs crowd behavior intelligent perception analysis on the target population based on the overhead view thermal analysis diagram. Compared with the traditional crowd behavior analysis method, since the above method of this embodiment performs an inverse perspective transformation on the conventional thermal analysis diagram of the target population to obtain an overhead view thermal analysis diagram, it is possible to perform crowd behavior intelligent perception analysis on the target population based on the overhead view thermal analysis diagram, so that the user can intuitively analyze various types of crowd behaviors from an overhead view perspective, and can use the same video file to perform spatial division or temporal division on the same space to comprehensively study the relationship between crowd activities and space in the space.

[0094] Based on the first embodiment of the intelligent perception and analysis device of crowd behavior based on computer vision of the present application, a second embodiment of the intelligent perception and analysis device of crowd behavior based on computer vision of the present application is proposed.

[0095] In this embodiment, the data acquisition module 601 is also used to scan and identify the target video file frame by frame through the YOLOX algorithm and the ByteTrack algorithm to obtain a number of video frames; color and number the video frames to obtain the crowd distribution data corresponding to the target population.

[0096] Furthermore, the tracking and identification module 602 is also used to perform regional screening of the crowd distribution data based on preset analysis targets to obtain screened crowd distribution data; and to track and identify the target population according to the crowd information and crowd behavior corresponding to the target population in the screened crowd distribution data to obtain crowd behavior data.

[0097] Furthermore, the conventional thermal analysis diagram includes a crowd distance distribution thermal analysis diagram, a crowd occupancy rate thermal analysis diagram, a crowd density distribution thermal analysis diagram and a crowd trajectory distribution thermal analysis diagram. The image generation module 603 is also used to generate a crowd distance distribution thermal analysis diagram corresponding to the target crowd based on the crowd distance, and the crowd distance is obtained by tracking and identifying the pedestrian's footsteps frame by frame; and to generate a crowd occupancy rate thermal analysis diagram corresponding to the target crowd based on the crowd residence time, and the crowd residence time is the time that the pedestrians tracked and identified at the same point in the target video file stay.

[0098] Furthermore, the image generation module 603 is also used to generate a crowd density distribution heat map corresponding to the target population based on the crowd density, wherein the crowd density is the number of pedestrians passing the same point in the target video file; and to generate a crowd trajectory distribution heat map corresponding to the target population based on the crowd trajectory, wherein the crowd trajectory is the movement trajectory of the pedestrians in the target video file.

[0099] Furthermore, the transformation analysis module 604 is also used to take any rectangular area in the target video file as a perspective reference, and perform an inverse perspective transformation on the conventional thermal analysis diagram based on the perspective reference to obtain a top-down thermal analysis diagram; based on the top-down thermal analysis diagram, perform intelligent perception analysis of crowd behavior on the target population to obtain the distribution of human flow vitality in the area where the target population is located.

[0100] Other embodiments or specific implementation methods of the computer vision-based crowd behavior intelligent perception and analysis device of the present application can refer to the above-mentioned method embodiments and will not be repeated here.

[0101] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0102] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0103] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory / random access memory, a disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0104] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for intelligent perception and analysis of crowd behavior based on computer vision, characterized in that: The method comprises the following steps: Obtaining a target video file corresponding to a target population, and identifying crowd distribution data of the target population from the target video file; Tracking and identifying the target crowd according to the crowd flow distribution data to obtain crowd behavior data, wherein the crowd behavior data includes crowd distance, crowd residence time, crowd density and crowd trajectory; Generate a conventional thermodynamic analysis diagram corresponding to the target population based on the population behavior data; The conventional thermodynamic analysis diagram is subjected to an inverse perspective transformation to obtain an overhead view thermodynamic analysis diagram, and crowd behavior intelligent perception analysis is performed on the target population based on the overhead view thermodynamic analysis diagram.

2. The method for intelligent perception and analysis of crowd behavior based on computer vision according to claim 1, characterized in that: The step of identifying the crowd distribution data of the target group from the target video file includes: The target video file is scanned and identified frame by frame by using the YOLOX algorithm and the ByteTrack algorithm to obtain a number of video frames; The plurality of video frames are color-labeled and number-labeled to obtain crowd flow distribution data corresponding to the target population.

3. The method for intelligent perception and analysis of crowd behavior based on computer vision according to claim 1, characterized in that: The step of tracking and identifying the target crowd according to the crowd flow distribution data to obtain crowd behavior data includes: Performing regional screening on the crowd flow distribution data based on a preset analysis target to obtain screened crowd flow distribution data; The target population is tracked and identified according to the crowd information and crowd behavior corresponding to the target population in the filtered crowd flow distribution data to obtain crowd behavior data.

4. The method for intelligent perception and analysis of crowd behavior based on computer vision according to claim 1, characterized in that: The conventional thermodynamic analysis graph includes a crowd distance distribution thermodynamic graph, a crowd occupancy thermodynamic analysis graph, a crowd density distribution thermodynamic graph, and a crowd trajectory distribution thermodynamic graph. The step of generating the conventional thermodynamic analysis graph corresponding to the target crowd based on the crowd behavior data includes: Generate a crowd distance distribution heat map corresponding to the target crowd based on the crowd distance, wherein the crowd distance is obtained by tracking the pedestrian landing data identified frame by frame; A crowd occupancy rate heat analysis diagram corresponding to the target crowd is generated based on the crowd residence time, where the crowd residence time is the time that pedestrians tracked and identified at the same point in the target video file stay.

5. The method for intelligent perception and analysis of crowd behavior based on computer vision according to claim 4, characterized in that: The step of generating a conventional thermodynamic analysis diagram corresponding to the target population based on the population behavior data further includes: Generate a crowd density distribution heat map corresponding to the target crowd based on the crowd density, where the crowd density is the number of pedestrians passing by the same point in the target video file; A crowd trajectory distribution heat map corresponding to the target crowd is generated based on the crowd trajectory, where the crowd trajectory is the movement trajectory of pedestrians in the target video file.

6. The method for intelligent perception and analysis of crowd behavior based on computer vision according to claim 1, characterized in that: The step of performing an inverse perspective transformation on the conventional thermodynamic analysis diagram to obtain a top-view thermodynamic analysis diagram, and performing crowd behavior intelligent perception analysis on the target population based on the top-view thermodynamic analysis diagram, comprises: Taking any rectangular area in the target video file as a perspective reference, and performing an inverse perspective transformation on the conventional thermal analysis diagram based on the perspective reference to obtain a top-down thermal analysis diagram; Based on the top-view thermal analysis diagram, intelligent crowd behavior perception analysis is performed on the target population to obtain the crowd flow vitality distribution in the area where the target population is located.

7. A crowd behavior intelligent perception and analysis device based on computer vision, characterized in that: The crowd behavior intelligent perception analysis device based on computer vision comprises: A data acquisition module, used to acquire a target video file corresponding to a target group of people, and identify crowd distribution data of the target group of people from the target video file; A tracking and identification module is used to track and identify the target crowd according to the crowd flow distribution data to obtain crowd behavior data, wherein the crowd behavior data includes crowd distance, crowd residence time, crowd density and crowd trajectory; An image generation module, used for generating a conventional thermodynamic analysis diagram corresponding to the target population based on the population behavior data; The transformation analysis module is used to perform an inverse perspective transformation on the conventional thermal analysis diagram to obtain a top-view thermal analysis diagram, and perform intelligent perception analysis of crowd behavior on the target population based on the top-view thermal analysis diagram.

8. A crowd behavior intelligent perception and analysis device based on computer vision, characterized in that: The device includes: a memory, a processor, and a computer vision-based intelligent perception analysis program for crowd behavior stored in the memory and executable on the processor, wherein the computer vision-based intelligent perception analysis program for crowd behavior is configured to implement the steps of the computer vision-based intelligent perception analysis method for crowd behavior as described in any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer vision-based intelligent perception analysis program for crowd behavior is stored on the storage medium. When the computer vision-based intelligent perception analysis program for crowd behavior is executed by a processor, the steps of the computer vision-based intelligent perception analysis method for crowd behavior as described in any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product includes a computer vision-based intelligent perception analysis program for crowd behavior, which, when executed by a processor, implements the steps of the computer vision-based intelligent perception analysis method for crowd behavior as described in any one of claims 1 to 6.

Images