Pedestrian traffic behavior recognition system and method based on posture estimation

Abstract

A pedestrian traffic behavior recognition system and method based on posture estimation relates to the field of big data recognition, including: an acquisition module, an analysis module, an output module and a server. Based on posture estimation technology, posture key points are combined with torso vectors to match scattered key points into independent people in a bottom-up matching manner. The posture information of an indefinite number of people within the field of view can be recognized. The function of obtaining pedestrian posture coordinates by a posture estimation network is used to obtain the posture coordinates of pedestrians over a period of time in a large amount of traffic video data, and the current behavior is marked. A behavior recognition network with time series analysis capabilities is trained to realize the recognition of pedestrian behavior, and the prediction capability of the time series analysis network is used to predict the behavior of pedestrians to a certain extent in advance, so as to more timely discover and warn traffic violations, thereby solving the problem of the lack of understanding and prediction capabilities of pedestrian behavior in current visual target detection solutions.

Description

Technical Field

[0001] The present invention relates to the technical field of big data recognition, and in particular to a system and method for identifying pedestrian traffic behavior based on posture estimation. Background Art

[0002] In the past, infrared sensors were usually used to detect pedestrians' behaviors such as running red lights and occupying lanes, such as whether they had crossed the boundary or left the sidewalk, to determine whether the pedestrians had committed traffic violations. The existing visual solutions generally use target detection and face recognition technology to locate the coordinates of pedestrians and determine whether they are in illegal areas to detect violations. In current technologies, sensors have high requirements for infrastructure construction and have a limited range of action. The behaviors that can be detected are also limited by the functions of the sensors and are not easy to update.

[0003] Although the visual target detection solution has a wider range and can detect multiple people at the same time, it lacks the ability to understand and predict pedestrian behavior, such as whether pedestrians are playing with their mobile phones, walking normally, and whether they are about to commit illegal acts. Summary of the invention

[0004] The embodiments of the present invention provide a pedestrian traffic behavior recognition system and method based on posture estimation. Based on the posture estimation technology, the system uses posture key points combined with torso vectors to match scattered key points into independent people in a bottom-up matching manner. The system can identify the posture information of an indefinite number of people within the field of view. The system uses the function of obtaining the posture coordinates of pedestrians through a posture estimation network to obtain the posture coordinates of pedestrians over a period of time from a large amount of traffic video data, annotate the current behavior, and train a behavior recognition network with time series analysis capabilities to realize the recognition of pedestrian behavior. In addition, the system uses the prediction capability of the time series analysis network to predict the behavior of pedestrians to a certain extent in advance, thereby more timely discovering and warning traffic violations, and solving the problem of the current visual target detection scheme lacking the ability to understand and predict pedestrian behavior.

[0005] A pedestrian traffic behavior recognition system based on posture estimation, comprising: an acquisition module, an analysis module, an output module and a server;

[0006] The acquisition module is used to collect the image data of the pedestrian, process the image data and transmit the processed image data to the analysis module;

[0007] An analysis module, used for receiving the image data transmitted by the acquisition module, analyzing the image, and transmitting the analysis result data to the output module;

[0008] An output module, used to receive the result data transmitted by the analysis module, screen the result data, warn pedestrians who have traffic violations, and transmit the pedestrian traffic violation data to a server;

[0009] The server is used to receive and store the traffic violation data transmitted by the output module.

[0010] Furthermore, the acquisition module includes an acquisition unit and a preprocessing unit, the acquisition unit is used to acquire image data of pedestrians and transmit the image data to the preprocessing unit, and the preprocessing unit is used to receive the image data transmitted by the acquisition unit and preprocess the image data.

[0011] Furthermore, the preprocessing unit preprocesses the image data by dividing the image data into continuous picture frames, and performing image normalization and scaling processing to obtain image data of uniform specifications.

[0012] Furthermore, the analysis module includes a posture estimation network unit and a behavior recognition network unit. The posture estimation network unit is used to receive the image data transmitted by the acquisition module, analyze the image data to obtain pedestrian posture information, and transmit the obtained pedestrian posture information to the behavior recognition network unit. The behavior recognition network unit is used to receive the pedestrian posture information transmitted by the posture estimation network unit and recognize it and transmit the recognition result data to the output module.

[0013] Furthermore, the posture estimation network unit includes an analysis unit, a posture key point coordinate acquisition unit, a trunk prediction unit and a combination unit, wherein the analysis unit is used to receive the image data transmitted by the output module, and obtain the key point Gaussian feature map and the trunk vector feature map, and transmit the obtained key point Gaussian feature map and the trunk vector feature map to the posture key point coordinate acquisition unit and the trunk prediction unit respectively, the posture key point coordinate acquisition unit is used to receive the key point Gaussian feature map transmitted by the analysis unit and analyze it to obtain the posture key point coordinates, and transmit the obtained posture key point coordinates to the combination unit, the trunk prediction unit is used to receive the trunk vector feature map transmitted by the analysis unit and analyze it to obtain all trunk data, and transmit the obtained trunk data to the combination unit, the combination unit respectively receives the posture key point coordinates transmitted by the posture key point coordinate acquisition unit and the trunk data transmitted by the trunk prediction unit, connects all the obtained posture key point coordinates and the trunk data to obtain the pedestrian's posture information data, and splices all the posture information data for multiple consecutive frames to form temporally continuous posture information.

[0014] Furthermore, the output module includes a screening unit, a warning unit and an output unit. The screening unit is used to receive the recognition result data transmitted by the analysis module, and screen the recognition result data to obtain pedestrian traffic violation data, and transmit the pedestrian traffic violation data to the warning unit and the output unit respectively. The warning unit is used to receive the pedestrian traffic violation data transmitted by the screening unit and issue a warning message to the pedestrian. The output unit is used to receive the pedestrian traffic violation data transmitted by the screening unit and transmit the pedestrian traffic violation data to the server for storage.

[0015] In a second aspect, an embodiment of the present invention provides a method for identifying pedestrian traffic behavior based on posture estimation, comprising the following steps:

[0016] S1, obtaining pedestrian image data. The acquisition unit acquires the pedestrian image data and transmits the image data to the preprocessing unit. The preprocessing unit receives the image data transmitted by the acquisition unit, divides the image data into continuous picture frames, and performs image normalization and scaling processing to obtain image data of uniform specifications, and transmits the image data to the posture estimation network unit;

[0017] S2, image data analysis, the posture estimation network unit receives the image data transmitted by the preprocessing unit, analyzes the image data to obtain pedestrian posture information, and transmits the obtained pedestrian posture information to the behavior recognition network unit. The behavior recognition network unit receives the pedestrian posture information transmitted by the posture estimation network unit, recognizes it, and transmits the recognition result data to the screening unit;

[0018] S3, result output, the screening unit receives the recognition result data transmitted by the analysis module, and screens the recognition result data to obtain pedestrian traffic violation data, and transmits the pedestrian traffic violation data to the warning unit and the output unit respectively, the warning unit receives the pedestrian traffic violation data transmitted by the screening unit and issues a warning message to the pedestrian, the output unit receives the pedestrian traffic violation data transmitted by the screening unit and transmits the pedestrian traffic violation data to the server for storage.

[0019] The beneficial effects of the above technical solution provided by the embodiment of the present invention include at least:

[0020] The present invention is based on posture estimation technology, uses posture key points combined with torso vectors, and matches scattered key points into independent people in a bottom-up matching manner. It can identify the posture information of an indefinite number of people within the field of view, and utilizes the function of obtaining pedestrian posture coordinates through a posture estimation network. It can not only obtain the current coordinates of the pedestrian's posture within a period of time from a large amount of traffic video data, and mark the current behavior, train a behavior recognition network with time series analysis capabilities, but also obtain the pedestrian's motion information, such as whether he is playing with a mobile phone, which can be judged by the movements of the hands and head. On this basis, combined with multiple consecutive posture information on the timeline, the pedestrian's upcoming behavior can be predicted through the pedestrian's current motion trend, thereby realizing the recognition of pedestrian behavior, and utilizing the prediction capability of the time series analysis network, it is also possible to predict the pedestrian's behavior to a certain extent in advance, thereby more timely discovering and warning of traffic violations.

[0021] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings.

[0022] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0024] Figure 1 A schematic diagram of the structure of a pedestrian traffic behavior recognition system based on posture estimation disclosed in an embodiment of the present invention;

[0025] Figure 2 A schematic diagram of a flow chart of a method for identifying pedestrian traffic behavior based on posture estimation disclosed in an embodiment of the present invention;

[0026] Figure 3 It is a key point Gaussian feature map disclosed in the prior art;

[0027] Figure 4 is a torso vector feature map disclosed in the prior art;

[0028] Figure 5 This is a schematic diagram of pedestrian posture estimation disclosed in the prior art.

[0029] Reference numerals:

[0030] 100-acquisition module; 101-acquisition unit; 102-preprocessing unit; 200-analysis module; 201-posture estimation network unit; 2011-analysis unit; 2012-posture key point coordinate acquisition unit; 2013-trunk prediction unit; 2014-combination unit; 202-behavior recognition network unit; 300-output module; 301-screening unit; 302-warning unit; 303-output unit; 400-server. Specific embodiments

[0031] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0032] Embodiment 1

[0033] like Figure 1 and 3 -5, an embodiment of the present invention provides a pedestrian traffic behavior recognition system based on posture estimation, including: an acquisition module 100, an analysis module 200, an output module 300 and a server 400;

[0034] The acquisition module 100 collects image data of pedestrians, processes the image data and transmits the processed image data to the analysis module 200. The acquisition module 100 includes a collection unit 101 and a preprocessing unit 102. The collection unit 101 collects image data of pedestrians and transmits the image data to the preprocessing unit 102. The preprocessing unit 102 receives the image data transmitted by the collection unit 101 and preprocesses the image data. The preprocessing unit 102 preprocesses the image data by dividing the image data into continuous picture frames, and performing image normalization and scaling processing to obtain image data of uniform specifications.

[0035] Specifically, the acquisition unit 101 is set at each traffic intersection and various places where pedestrian traffic violations are likely to occur to collect image data of pedestrians, and transmit the image data to the preprocessing unit 102. The preprocessing unit 102 divides the image data into continuous picture frames, and performs image normalization and scaling processing to obtain image data of uniform specifications.

[0036] The analysis module 200 receives the image data transmitted by the acquisition module 100, analyzes the image, and transmits the analysis result data to the output module 300. The analysis module 200 includes a posture estimation network unit 201 and a behavior recognition network unit 202. The posture estimation network unit 201 receives the image data transmitted by the acquisition module 100, analyzes the image data to obtain pedestrian posture information, and transmits the obtained pedestrian posture information to the behavior recognition network unit 202. The posture estimation network unit 201 includes an analysis unit 2011, a posture key point coordinate acquisition unit 2012, a trunk prediction unit 2013 and a combination unit 2014. The analysis unit 2011 receives the image data transmitted by the output module 300, and obtains a key point Gaussian feature map and a trunk vector feature map, and transmits the obtained key point Gaussian feature map and the trunk vector feature map to the posture key point coordinate acquisition unit 2012 and the trunk prediction unit 2013 respectively. 2012 receives the key point Gaussian feature map transmitted by the analysis unit 2011 and analyzes it to obtain the coordinates of the key points of the posture, and transmits the obtained coordinates of the key points of the posture to the combination unit 2014. The trunk prediction unit 2013 receives the trunk vector feature map transmitted by the analysis unit 2011 and analyzes it to obtain all the trunk data, and transmits the obtained trunk data to the combination unit 2014. The combination unit 2014 receives the posture key point coordinates transmitted by the posture key point coordinate acquisition unit 2012 and the trunk data transmitted by the trunk prediction unit 2013 respectively, connects all the obtained posture key point coordinates and trunk data to obtain the posture information data of the pedestrian, and splices all the posture information data for multiple consecutive frames to form temporally continuous posture information, and transmits the posture information to the recognition network unit. The behavior recognition network unit 202 receives the pedestrian posture information transmitted by the posture estimation network unit 201, recognizes it, and transmits the recognition result data to the output module 300.

[0037] Specifically, Figure 3-4 As shown, the image data is input to the analysis unit 2011 to obtain N key point Gaussian feature maps and M trunk vector feature maps. The posture key point coordinate acquisition unit 2012 extracts the local maximum value of the N Gaussian feature maps to obtain the posture key point coordinates. The local extreme value calculation formula is as follows:

[0038] X∈R,||XX * ||<ε

[0039] ε>0

[0040] f(X)≥f(X * )

[0041] Among them, X represents any point in the feature map, X * Represents the coordinate point near X, ε is used to specify X *The distance from X, f(X) is the value corresponding to the coordinate point;

[0042] The analysis unit 2011 predicts the trunk in the form of a vector, and predicts the unit vector of the trunk direction, that is, M trunk vector feature maps. Each coordinate point of the feature map is a vector value (x, y), and there is no vector value (x, y)→(0, 0) between the key points of the trunk. Using this feature, the trunk prediction unit 2013 generates a unit vector with every two key points as the connecting direction, and performs dot product with all the vectors of the trunk vector feature map on the path and sums them. The key point pairs with the dot product sum higher than the threshold are selected to find all the trunks. The combination unit 2014 connects all the key points and their trunks to form an undirected graph, and uses the binary matching Hungarian algorithm to calculate the maximum match and combine them into independent people, so as to obtain the posture information data of all detected pedestrians. The posture information of each person is spliced ​​for multiple consecutive frames to form temporally continuous posture information. The temporal posture information of each person is normalized and then input into the behavior recognition network unit 202 in sequence to predict the pedestrian behavior, and the prediction result data is transmitted to the screening unit 301.

[0043] In this embodiment, the posture coordinates of pedestrians over a period of time are obtained from a large amount of traffic video data, and the current behavior is marked and input into the behavior recognition network unit 202 to train the behavior recognition network unit 202 with time series analysis capabilities.

[0044] The output module 300 receives the result data transmitted by the analysis module 200, screens the result data, warns pedestrians who have traffic violations, and transmits the pedestrian traffic violation data to the server 400. The output module 300 includes a screening unit 301, a warning unit 302, and an output unit 303. The screening unit 301 receives the recognition result data transmitted by the analysis module 200, screens the recognition result data, obtains pedestrian traffic violation data, and transmits the pedestrian traffic violation data to the warning unit 302 and the output unit 303 respectively. The warning unit 302 receives the pedestrian traffic violation data transmitted by the screening unit 301 and issues a warning message to the pedestrian. The output unit 303 receives the pedestrian traffic violation data transmitted by the screening unit 301 and transmits the pedestrian traffic violation data to the server 400 for storage.

[0045] Specifically, the screening unit 301 receives the prediction result data transmitted by the behavior recognition network unit 202, screens the prediction result data, and screens out the data of pedestrian traffic violations. For example, if a pedestrian runs a red light, the screening unit 301 screens out the pedestrian's red light running behavior, and transmits the pedestrian traffic violation data to the warning unit 302 and the output unit 303 respectively. After receiving the traffic violation data, the warning unit 302 warns the pedestrian, for example, through sound and light reminders. The output unit 303 receives the pedestrian traffic violation behavior data transmitted by the screening unit 301 and transmits the pedestrian traffic violation behavior data to the server 400 for storage.

[0046] The present invention is based on posture estimation technology, uses posture key points combined with torso vectors, and matches scattered key points into independent people in a bottom-up matching manner. It can identify the posture information of an indefinite number of people within the field of view, and utilizes the function of obtaining pedestrian posture coordinates by the posture estimation network unit 201. It can not only obtain the current coordinates of the pedestrian's posture within a period of time from a large amount of traffic video data, and mark the current behavior, and train the behavior recognition network unit 202 with time series analysis capabilities, but also obtain the pedestrian's motion information, such as whether to play with a mobile phone, which can be judged by the movements of the hands and head. On this basis, combined with multiple consecutive posture information on the timeline, the pedestrian's upcoming behavior can be predicted through the pedestrian's current movement trend, thereby realizing the recognition of pedestrian behavior, and utilizing the prediction ability of the time series analysis network, it is also possible to predict the pedestrian's behavior to a certain extent in advance, thereby more timely discovering and warning traffic violations.

[0047] Embodiment 2

[0048] The embodiment of the present invention also discloses a method for identifying pedestrian traffic behavior based on posture estimation, such as Figure 2 , including the following steps:

[0049] S1, obtaining pedestrian image data, the acquisition unit 101 acquires the pedestrian image data, and transmits the image data to the preprocessing unit 102, the preprocessing unit 102 receives the image data transmitted by the acquisition unit 101, divides the image data into continuous picture frames, and performs image normalization and scaling processing to obtain image data of uniform specifications, and transmits the image data to the posture estimation network unit 201;

[0050] Specifically, the preprocessing unit 102 divides the image data collected by the collecting unit 101 into continuous picture frames, and performs image normalization and scaling processing to obtain image data of uniform specifications, so as to facilitate subsequent processing.

[0051] S2, image data analysis, the posture estimation network unit 201 receives the image data transmitted by the preprocessing unit 102, analyzes the image data to obtain pedestrian posture information, and transmits the obtained pedestrian posture information to the behavior recognition network unit 202, the behavior recognition network unit 202 receives the pedestrian posture information transmitted by the posture estimation network unit 201, recognizes it, and transmits the recognition result data to the screening unit 301;

[0052] S3, result output, the screening unit 301 receives the recognition result data transmitted by the analysis module 200, and screens the recognition result data to obtain pedestrian traffic violation data, and transmits the pedestrian traffic violation data to the warning unit 302 and the output unit 303 respectively, the warning unit 302 receives the pedestrian traffic violation data transmitted by the screening unit 301 and issues a warning message to the pedestrian, the output unit 303 receives the pedestrian traffic violation data transmitted by the screening unit 301 and transmits the pedestrian traffic violation data to the server 400 for storage.

[0053] The pedestrian traffic behavior recognition method based on posture estimation disclosed in this embodiment uses posture key points combined with torso vectors to match scattered key points into independent people in a bottom-up matching manner, and can identify the posture information of an indefinite number of people within the field of view. The function of obtaining pedestrian posture coordinates by a posture estimation network is used to obtain the posture coordinates of pedestrians over a period of time from a large amount of traffic video data, and the current behavior is marked to train a behavior recognition network with time series analysis capabilities, thereby realizing the recognition of pedestrian behavior. In addition, by using the prediction ability of the time series analysis network, it is possible to predict the behavior of pedestrians to a certain extent in advance, thereby solving the problem of the lack of understanding and prediction capabilities of pedestrian behavior in current visual target detection solutions.

[0054] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process can be rearranged without departing from the scope of protection of the present disclosure. The attached method claims present the elements of the various steps in an exemplary order and are not intended to be limited to the specific order or hierarchy described.

[0055] In the above detailed description, various features are grouped together in a single embodiment to simplify the disclosure. This method of disclosure should not be interpreted as reflecting an intention that the embodiments of the claimed subject matter require more features than are clearly stated in each claim. On the contrary, as reflected in the appended claims, the invention is in a state of having less than all the features of the disclosed individual embodiments. Therefore, the appended claims are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment of the invention.

[0056] Those skilled in the art will also appreciate that the various illustrative logic blocks, modules, circuits, and algorithmic steps described in conjunction with the embodiments herein can all be implemented as electronic hardware, computer software, or a combination thereof. In order to clearly illustrate the interchangeability between hardware and software, various illustrative components, blocks, modules, circuits, and steps are generally described above around their functions. Whether such functions are implemented as hardware or software depends on specific applications and the design constraints imposed on the entire system. A skilled person can implement the described functions in an alternative manner for each specific application, but such implementation decisions should not be interpreted as departing from the scope of protection of the present disclosure.

[0057] The steps of the method or algorithm described in conjunction with the embodiments herein may be directly embodied as hardware, a software module executed by a processor, or a combination thereof. The software module may be located in a RAM memory, a flash memory, a ROM memory, an EPROM memory, an EEPROM memory, a register, a hard disk, a mobile disk, a CD-ROM, or any other form of storage medium well known in the art. An exemplary storage medium is connected to the processor so that the processor can read information from the storage medium and can write information to the storage medium. Of course, the storage medium may also be an integral part of the processor. The processor and the storage medium may be located in an ASIC. The ASIC may be located in a user terminal. Of course, the processor and the storage medium may also be present in a user terminal as discrete components.

[0058] For software implementation, the techniques described in this application can be implemented with modules (e.g., procedures, functions, etc.) that perform the functions described in this application. These software codes can be stored in a memory unit and executed by a processor. The memory unit can be implemented within the processor or outside the processor. In the latter case, it is coupled to the processor in a communication manner via various means, which are well known in the art.

[0059] The above description includes examples of one or more embodiments. Of course, it is impossible to describe all possible combinations of components or methods for the purpose of describing the above embodiments, but it should be recognized by those skilled in the art that the various embodiments may be further combined and arranged. Therefore, the embodiments described herein are intended to cover all such changes, modifications and variations that fall within the scope of protection of the appended claims. In addition, with respect to the term "comprising" used in the specification or claims, the word is covered in a manner similar to the term "including", just as "including," is explained as a transitional word in the claims. In addition, any term "or" used in the specification of the claims is intended to mean "non-exclusive or".

Claims

1. A pedestrian traffic behavior recognition system based on posture estimation, characterized in that: include: Acquisition module, analysis module, output module and server; The acquisition module is used to collect the image data of the pedestrian, process the image data and transmit the processed image data to the analysis module; An analysis module, used for receiving the image data transmitted by the acquisition module, analyzing the image, and transmitting the analysis result data to the output module; The analysis module includes a posture estimation network unit and a behavior recognition network unit; The posture estimation network unit is used to receive the image data transmitted by the acquisition module, analyze the image data to obtain pedestrian posture information, and transmit the obtained pedestrian posture information to the behavior recognition network unit; Obtain the posture coordinates of pedestrians over a period of time from a large amount of traffic video data, annotate the current behavior, input them into the behavior recognition network unit, train the behavior recognition network unit with time series analysis capability, the behavior recognition network unit is used to receive the pedestrian posture information transmitted by the posture estimation network unit and recognize it, and transmit the recognition result data to the output module; Wherein, the posture estimation network unit includes an analysis unit, a posture key point coordinate acquisition unit, a torso prediction unit and a combination unit; The analysis unit is used to receive the image data transmitted by the output module, and obtain the key point Gaussian feature map and the trunk vector feature map, and transmit the obtained key point Gaussian feature map and the trunk vector feature map to the posture key point coordinate acquisition unit and the trunk prediction unit respectively; The posture key point coordinate acquisition unit is used to receive and analyze the key point Gaussian feature map transmitted by the analysis unit to obtain the posture key point coordinates, and transmit the obtained posture key point coordinates to the combination unit; The trunk prediction unit is used to receive and analyze the trunk vector feature map transmitted by the analysis unit, obtain all trunk data, and transmit the obtained trunk data to the combination unit; The combination unit receives the posture key point coordinates transmitted by the posture key point coordinate acquisition unit and the torso data transmitted by the torso prediction unit, respectively, connects all the posture key point coordinates and torso data obtained to obtain the posture information data of the pedestrian, and splices all the posture information data for multiple consecutive frames to form temporally continuous posture information; An output module, used to receive the result data transmitted by the analysis module, screen the result data, warn pedestrians who have traffic violations, and transmit the pedestrian traffic violation data to a server; The output module includes a screening unit, a warning unit and an output unit. The screening unit is used to receive the recognition result data transmitted by the analysis module, and screen the recognition result data to obtain pedestrian traffic violation data, and transmit the pedestrian traffic violation data to the warning unit and the output unit respectively. The warning unit is used to receive the pedestrian traffic violation data transmitted by the screening unit and issue a warning message to the pedestrian. The output unit is used to receive the pedestrian traffic violation data transmitted by the screening unit and transmit the pedestrian traffic violation data to the server for storage. The server is used to receive and store the traffic violation data transmitted by the output module.

2. The pedestrian traffic behavior recognition system based on posture estimation according to claim 1, characterized in that: The acquisition module includes an acquisition unit and a preprocessing unit. The acquisition unit is used to acquire image data of pedestrians and transmit the image data to the preprocessing unit. The preprocessing unit is used to receive the image data transmitted by the acquisition unit and preprocess the image data.

3. The pedestrian traffic behavior recognition system based on posture estimation as claimed in claim 2, characterized in that: The preprocessing unit preprocesses the image data by dividing the image data into continuous picture frames, and performing image normalization and scaling processing to obtain image data of uniform specifications.

4. A method for identifying pedestrian traffic behavior based on posture estimation, applied to the pedestrian traffic behavior identification system based on posture estimation according to claim 1, characterized in that: The following steps are involved: S1, obtaining pedestrian image data. The acquisition unit acquires the pedestrian image data and transmits the image data to the preprocessing unit. The preprocessing unit receives the image data transmitted by the acquisition unit, divides the image data into continuous picture frames, and performs image normalization and scaling processing to obtain image data of uniform specifications, and transmits the image data to the posture estimation network unit; S2, image data analysis, the posture estimation network unit receives the image data transmitted by the preprocessing unit, analyzes the image data to obtain pedestrian posture information, and transmits the obtained pedestrian posture information to the behavior recognition network unit. The behavior recognition network unit receives the pedestrian posture information transmitted by the posture estimation network unit, recognizes it, and transmits the recognition result data to the screening unit; S3, result output, the screening unit receives the recognition result data transmitted by the analysis module, and screens the recognition result data to obtain pedestrian traffic violation data, and transmits the pedestrian traffic violation data to the warning unit and the output unit respectively, the warning unit receives the pedestrian traffic violation data transmitted by the screening unit and issues a warning message to the pedestrian, the output unit receives the pedestrian traffic violation data transmitted by the screening unit and transmits the pedestrian traffic violation data to the server for storage.

Images