Method, system, electronic device and storage medium for detecting theft in public places

Abstract

The application discloses a public place thief detection method and system, electronic equipment and storage medium, the public place thief detection method comprises the following steps: detecting people through a human key point detection model, and calculating a human body frame through human key points; calculating the intersection and union ratio between the human body frames, and cutting out the pictures of the positions between the thighs and necks of the two people through the position information of the human hand key points; inputting the cut-out pictures of the positions between the thighs and necks of the two people into a Cls-UNet model to obtain segmentation results and classification results; and in the case that the segmentation results and the classification results both satisfy preset conditions and the front and back frame relationships satisfy preset conditions, it is determined that a theft behavior occurs. The detection method of the application adopts a semantic segmentation UNet network Cls-Unet with a classification function to analyze images, performs image segmentation and classification on the cut-out pictures, realizes theft behavior detection, has a wide application range, can be applied to large space environments, and has clear shooting and low cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robot detection, and more particularly to a public place thief detection method and system, an electronic device and a storage medium. BACKGROUND

[0002] At present, in public places such as train station waiting hall, bus station waiting hall, bus station, bus, large shopping mall, market, square, etc., small robot hands with sensors are mainly used to prevent theft, convex mirrors are installed in buses to prevent theft, or a large number of cameras are installed to monitor or for subsequent evidence collection, so as to realize the detection of thief behavior.

[0003] However, in the above method of the prior art, the small robot hand has limited application range and is more suitable for use as a toy. The detection method of installing a large number of convex mirrors is not suitable for large space environment, and the whole does not meet the actual use requirement. The camera monitoring picture is generally far away from the crowd, and many cases are not clear. In addition, the camera is basically fixed, and the cost is high if a large number of cameras are installed. SUMMARY

[0004] An object of the present application is to provide a new technical solution of a public place thief detection method, system, electronic device and storage medium, which can at least solve the problems of limited application range, inapplicability to large space environment, unclear shooting and high cost in the prior art.

[0005] In a first aspect, the present application provides a public place thief detection method, comprising:

[0006] detecting a person through a human key point detection model, and calculating a human body frame through the human key point;

[0007] calculating the intersection-over-union of the human body frames, and cropping a picture of the part between the thighs and the neck of the two persons through the position information of the human hand key points;

[0008] inputting the cropped picture of the part between the thighs and the neck of the two persons into a Cls-UNet model to obtain a segmentation result and a classification result, and determining that a theft behavior occurs when the segmentation result and the classification result both satisfy a preset condition and the front and back frame relationship satisfies a preset condition.

[0009] Optionally, the public place thief detection method further comprises:

[0010] cropping pictures of the head, upper body and lower body corresponding to the position information of the human key points;

[0011] determining whether a hat is worn through a classification model, and analyzing the color of clothes and trousers through HSV color space;

[0012] According to the judgment result of the classification model and the analysis result of the HSV color space, an alarm sound is issued to remind the stolen person.

[0013] Optionally, the public place thief detection method further comprises: storing a video when the theft behavior occurs as evidence for evidence collection.

[0014] Optionally, the step of calculating the intersection over union between the human body frames comprises:

[0015] The video frames of the current frame, the first N frames before the current frame and the first N frames after the current frame are stored.

[0016] After the first N frames after the first N frames, the intersection over union of the first N frames is calculated.

[0017] Optionally, the formula for calculating the intersection over union between the human body frames is:

[0018]

[0019] Wherein, IoU represents the intersection over union, A and B represent two human body frames respectively.

[0020] The relationship formula of the preset condition satisfied by the relationship between the front and rear frames is:

[0021]

[0022] When the IoU of the first N frames before the current frame is equal to 0, the IoU of the current frame is greater than 0, and the IoU of the first N frames after the current frame is equal to 0, it is determined that the theft behavior occurs. t’ t t”

[0023] Optionally, the segmentation categories of the Cls-UNet model include hands, backpacks, single-shoulder bags, mobile phones and wallets, and the classification number of the classification is 2, wherein 0 represents not stealing things, and 1 represents stealing things.

[0024] Optionally, when the Cls-UNet model is segmented to hands, and any one of a backpack, a single-shoulder bag, a mobile phone and a wallet, and the classification category channel output of the Cls-UNet model is 1, it is determined that the theft behavior occurs.

[0025] In a second aspect of the present application, a public place thief detection system is provided, which is applied to the public place thief detection method described in the above embodiments, and the detection system comprises:

[0026] A detection module detects a person through a human key point detection model.

[0027] ​​​The first calculation module calculates a human body frame through the human body key points;

[0028] The second calculation module is used for calculating the intersection-over-union between the human body frames;

[0029] The cropping module crops a picture of a part between the two thighs and the neck through the position information of the human body hand key points;

[0030] The logical analysis module is used for inputting the cropped picture of the part between the two thighs and the neck into a Cls-UNet model to obtain a segmentation result and a classification result, and in the case that the segmentation result and the classification result both satisfy preset conditions and a front-back frame relationship satisfies a preset condition, it is determined that a theft behavior occurs.

[0031] In a third aspect, the present application provides an electronic device, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are run by the processor, the processor executes the steps of the public place thief detection method in the above embodiments.

[0032] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and when the computer program is run by a processor, the processor executes the steps of the public place thief detection method in the above embodiments.

[0033] The public place thief detection method of the present application uses a patrol robot to detect human body key points, and the patrol robot can penetrate into a crowd and capture clearer images than fixed cameras. Moreover, the patrol robot can automatically move, thereby saving the cost of cameras. The human body frame and the intersection-over-union between the human body frames are calculated through the human body key points, and a semantic segmentation UNet network Cls-Unet with a classification function is used for image analysis, and the cropped picture is subjected to image segmentation and classification, thereby realizing theft behavior detection. The public place thief detection method has a wide application range, can be applied to large space environments, has clear images and low cost, and is suitable for large-area popularization and use.

[0034] Other features and advantages of the present application will become apparent from the following detailed description of exemplary embodiments thereof, taken in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0035] The accompanying drawings incorporated in and forming a part of the specification, illustrate embodiments of the present application and, together with the description, serve to explain the principles of the application.

[0036] Figure 1is a flow chart of a public place thief detection method according to an embodiment of the present application;

[0037] Figure 2 is another flow chart of a public place thief detection method according to an embodiment of the present application;

[0038] Figure 3 is a working principle diagram of an electronic device according to an embodiment of the present application.

[0039] Reference Signs:

[0040] a processor 201;

[0041] a memory 202; an operating system 2021; an application program 2022;

[0042] a network interface 203;

[0043] an input device 204;

[0044] a hard disk 205;

[0045] a display device 206. DETAILED DESCRIPTION

[0046] Various exemplary embodiments of the present application will now be described in detail with reference to the accompanying drawings. It should be noted that the relative arrangement of the components and steps set forth in the examples, numerical expressions, and numerical values set forth in the examples are not limiting to the scope of the present application unless otherwise specifically stated.

[0047] The following description of at least one exemplary embodiment is merely exemplary in nature and is in no way intended to limit the scope of the application, its application, or uses.

[0048] Techniques, methods, and devices known to those of ordinary skill in the relevant art can not be discussed in detail herein. However, the techniques, methods, and devices are sufficiently described in the disclosure.

[0049] In all of the examples shown and discussed herein, any specific values should be interpreted as merely exemplary and not as a limitation. Thus, other examples of the exemplary embodiments can have different values.

[0050] It should be noted that like reference numerals and letters refer to like items in the following drawings, and thus, once an item is defined in one drawing, it is not necessary to discuss it further in subsequent drawings.

[0051] In the description and claims of the application, the terms "first", "second", etc. can include one or more of the features can be explicitly or implicitly included. In the description of the application, unless otherwise specified, the meaning of "a plurality of" is two or more. In addition, "and / or" in the specification and claims means at least one of the connected objects, and the character " / ", generally indicates that the front and rear associated objects are in a "or" relationship.

[0052] In the description of the application, it should be understood that, if the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, only for the convenience of describing the application and simplifying the description, and not to indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, therefore cannot be understood as a limitation of the application.

[0053] In the description of the application, it should be noted that, unless otherwise specified and limited, the terms "mounting", "connecting", "connecting" should be understood broadly. For example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium; it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the application can be understood according to the specific circumstances.

[0054] The public place thief detection method according to the embodiment of the application will be described in detail below with reference to the drawings.

[0055] The public place thief detection method according to the embodiment of the application comprises:

[0056] S1, detecting people through a human key point detection model, and calculating a human body frame through human key points;

[0057] S2, calculating the intersection and union ratio between the human body frames, and cutting out the pictures of the parts between the thighs and necks of the two people through the position information of the human hand key points;

[0058] S3, inputting the cut-out pictures of the parts between the thighs and necks of the two people into a Cls-UNet model to obtain segmentation results and classification results, and determining that a theft behavior occurs when the segmentation results and the classification results both satisfy the preset conditions, and the front and rear frame relationship satisfies the preset conditions.

[0059] In other words, the public place thief detection method according to the embodiment of the present application can realize the function of shooting thieves in public places through the patrol robot and by using image analysis and other technologies. In the public place thief detection method of the present application, first, the patrol robot can be used to detect people through a human key point detection model and calculate a human body frame through human key points. The patrol robot can penetrate into the crowd and capture clearer images than fixed cameras. Moreover, the patrol robot can automatically move, saving the cost of cameras.

[0060] Then, the intersection over union (IOU) between the human body frames can be calculated according to the calculated human body frames, and the pictures of the parts between the thighs and necks of the two people can be cropped through the position information of the human hand key points. The IOU is mainly used to describe the degree of overlap of two frames, and the larger the overlap area, the larger the value of the IOU.

[0061] Finally, the pictures of the parts between the thighs and necks of the two people can be input into a Cls-UNet model to perform image analysis by using a semantic segmentation UNet network Cls-Unet with classification function, to obtain segmentation results and classification results. In the case that the segmentation results and classification results both meet preset conditions and the front and back frame relationships meet preset conditions, it is determined that a theft behavior occurs. The public place thief detection method has a wide range of applications, can be applied to large space environments, and has clear shooting, low cost, and is suitable for large-area popularization and use.

[0062] Therefore, the public place thief detection method according to the embodiment of the present application uses the patrol robot to perform human key point detection, the patrol robot can penetrate into the crowd and capture clearer images than fixed cameras. Moreover, the patrol robot can automatically move, saving the cost of cameras. The human body frame and the IOU between the human body frames are calculated through the human key points, and image analysis is performed by using a semantic segmentation UNet network Cls-Unet with classification function, to perform image segmentation and classification on the cropped pictures, realizing theft behavior detection. The public place thief detection method has a wide range of applications, can be applied to large space environments, and has clear shooting, low cost, and is suitable for large-area popularization and use.

[0063] In some specific embodiments of the present application, the public place thief detection method further comprises:

[0064] The pictures of the corresponding head, upper body and lower body are cropped through the position information of the human key points;

[0065] Whether a hat is worn is determined through a classification model, and the colors of clothes and trousers are analyzed through an HSV color space;

[0066] According to the judgment result of the classification model and the analysis result of the HSV color space, an alarm sound is issued to remind the stolen person.

[0067] The video when the theft occurs is stored as evidence for evidence collection.

[0068] That is, in the public place thief detection method, the corresponding head, upper body and lower body pictures can also be cropped through the position information of the human key points. And whether to wear a hat is judged through a classification model, and the color of the clothes and trousers is analyzed through the HSV color space. In the present application, whether the person wears a hat, the head picture can be cropped by using the head key point detection result of the human key point, and a two-classifier of wearing a hat or not wearing a hat is trained. For the color of the clothes and trousers, the present application uses the key point results of the body and leg detected by the human key point to crop the upper body and lower body pictures, and the color of the clothes and trousers is identified by converting to the HSV color space. The color with the largest proportion is considered to be the color. Finally, according to the judgment result of the classification model and the analysis result of the HSV color space, an alarm sound is issued to remind the stolen person. At the same time, the video when the theft occurs is stored as evidence for evidence collection.

[0069] In order to timely remind the stolen person, the present application identifies whether the stolen person and the thief wear a hat, wear clothes of what color and wear trousers of what color, and then issues a sound to remind the stolen person, for example, "the student wearing white clothes and black trousers, your things have been stolen by a person wearing black clothes, black trousers and a hat". Because the thief will quickly walk away or run away after stealing things, the stolen person can quickly lock the thief from the crowd after hearing the sound, which can help the stolen person to stop the loss in time to a great extent.

[0070] In some specific embodiments of the present application, the step of calculating the intersection-over-union between the human body frames comprises:

[0071] The video frames of the current frame, the first N frames before the current frame and the first N frames after the current frame are stored.

[0072] After N frames after the first N frames, the intersection-over-union of the first N frames is calculated.

[0073] The formula for calculating the intersection-over-union between the human body frames is:

[0074]

[0075] Wherein, IoU represents the intersection-over-union, A and B represent two human body frames respectively.

[0076] The relationship formula of the preset condition satisfied by the relationship between the front and rear frames is:

[0077]

[0078] wherein, when the IoU of the first N frames t’ equals 0, the IoU of the current frame t is greater than 0, the IoU of the first N frames t” equals 0, it is determined that a theft occurs.

[0079] The segmentation categories of the Cls-UNet model include: hand, backpack, single-shoulder bag, mobile phone and wallet, and the classification number of the classification is 2, wherein 0 represents not stealing, and 1 represents stealing.

[0080] That is, it needs to be explained that through analysis, generally, a thief will first approach the target person, then steal some items from the pocket or backpack of the target person, and finally the thief leaves the target person, so the three key moments can be identified. The present application adopts the method of human key points to detect the human body, analyzes the second moment in detail, and logically analyzes the characteristics of the three moments, and finally identifies the thief.

[0081] In the process of calculating the intersection over union between the human body frames, the robot will automatically store the current frame, the first N frames of the current frame and the first N frames of the current frame. The video frame of the first N frames is the video frame at the current moment, which is the first N frames relative to the theft moment, which is equivalent to the thief stealing the things, and then after N frames, the intersection over union of the first N frames is calculated. The formula for calculating the intersection over union between the human body frames is:

[0082]

[0083] wherein, IoU represents the intersection over union, A and B represent two human body frames respectively. After meeting the conditions, an alarm is generated.

[0084] Suppose the video frequency of the robot camera is 30 FPS, then the N frame time consumption is N / 30 seconds, wherein N is 30 by default. When the IoU of the two human body frames is greater than 0, it is considered that a suspected theft occurs. Then, the upper body parts of the two are cropped through the key point position information of the hand, and whether a theft occurs is judged through a newly designed classifier.

[0085] The relationship formula of the preset condition satisfied by the front and back frame relationship is:

[0086]

[0087] wherein, when the IoU of the first N frames t’ equals 0, the IoU of the current frame t is greater than 0, the IoU of the first N frames t” equals 0, it is determined that a theft occurs.

[0088] If the classifier determines that a theft occurs, after 10 frames, the IoU of the two human target boxes is analyzed. If the IoU of the first N frames is equal to 0, the IoU of the current frame is greater than 0, and the IoU of the last N frames is equal to 0, as shown in the formula, it is determined that a theft occurs. t IoU represents the IoU of the two human target boxes relative to the current frame at the time of theft. t’ IoU represents the IoU of the two human target boxes relative to the first N frames at the time of theft. t” IoU represents the IoU of the two human target boxes relative to the last N frames at the time of theft.

[0089] In some embodiments of the present application, the segmentation categories of the Cls-UNet model include hands, backpacks, single-shoulder bags, mobile phones, and wallets, and the classification category number is 2, where 0 represents not stealing, and 1 represents stealing. When the segmentation is to the hands, and any one of the backpacks, single-shoulder bags, mobile phones, and wallets, and the classification category channel output of the Cls-UNet model is 1, it is determined that a theft occurs.

[0090] In other words, due to the special nature of theft, using a common classification network may produce more false positives, so the present application innovatively modifies a semantic segmentation network UNet to become a segmentation network with classification function, and the new network is named Cls-UNet, which can provide more information for judging theft. Compared with the ordinary UNet, the input of Cls-UNet is consistent with UNet, and the output increases a dimension to display its category. The number of segmentation categories is 5: hands, backpacks, single-shoulder bags, mobile phones, and wallets; the number of classification categories is 2: 0 and 1 (0 represents not stealing, and 1 represents stealing, and the proportion of a certain value in the category channel is greater than 80% before it is considered valid, and its category is this value). The final judgment logic is that the segmentation is to the hands, and any one of the backpacks, single-shoulder bags, mobile phones, and wallets, and the classification category channel output is 1, then it is determined that a theft occurs.

[0091] In summary, according to the public place thief detection method of the present application, the patrol robot is used for human key point detection, and the patrol robot can penetrate into the crowd and capture clearer images than fixed cameras. Moreover, the patrol robot can automatically move, saving the cost of cameras. The human frame and the intersection-over-union between human frames are calculated through human key points, and image analysis is performed using a semantic segmentation UNet network Cls-Unet with classification function to realize theft behavior detection. The public place thief detection method has a wide range of applications, can be applied to large space environments, and has clear images and low cost, making it suitable for large-scale popularization and use.

[0092] According to a second aspect of the present application, a public place thief detection system is provided, which is applied to the public place thief detection method in the above-mentioned embodiments. The detection system comprises a detection module, a first calculation module, a second calculation module, a cropping module and a logic analysis module. The detection module detects a person through a human key point detection model. The first calculation module calculates a human body frame through human key points. The second calculation module is used to calculate the intersection over union between human body frames. The cropping module crops pictures of the positions between the thighs and necks of two people through the position information of human hand key points. The logic analysis module is used to input the cropped pictures of the positions between the thighs and necks of two people into a Cls-UNet model to obtain segmentation results and classification results. In the case that the segmentation results and the classification results both satisfy preset conditions and the front and back frame relationships satisfy preset conditions, it is determined that a theft behavior occurs.

[0093] The public place thief detection system according to the embodiments of the present application detects human key points by using a patrol robot. The patrol robot can go deep into a crowd and can shoot clearer images than fixed cameras. Moreover, the patrol robot can automatically move, thereby saving the cost of cameras. The human body frame and the intersection over union between human body frames are calculated through human key points. Semantic segmentation UNet networks with classification functions such as Cls-Unet are used for image analysis. The cropped pictures are subjected to image segmentation and classification, thereby realizing theft behavior detection. The public place thief detection method has a wide application range, can be applied to large space environments, and has clear shooting, low cost and is suitable for large-area popularization and use.

[0094] According to a third aspect of the present application, an electronic device is further provided, which comprises a processor 201 and a memory 202, and the memory 202 stores computer program instructions. When the computer program instructions are run by the processor 201, the processor 201 executes the steps of the public place thief detection method in the above-mentioned embodiments.

[0095] Further, as shown in Figure 3 The electronic device further comprises a network interface 203, an input device 204, a hard disk 205 and a display device 206.

[0096] The various interfaces and devices described above can be interconnected through a bus architecture. The bus architecture can include any number of interconnecting buses and bridges. Various circuits can be connected to one or more central processing units (CPU) represented by the processor 201, along with various memories represented by the memory 202. The bus architecture also can include various other circuits such as peripheral devices, voltage regulators, and power management circuits. It is to be understood that the bus architecture is used to facilitate communication among these components and is not intended to limit the spirit or scope of the present application. In addition, various other circuits such as power on / off circuitry, reset circuitry, and clock circuits can be connected to the bus architecture.

[0097] The network interface 203 can be connected to a network (e.g., the Internet, a local area network, etc.) to obtain relevant data from the network and can store the data in the hard disk 205.

[0098] The input device 204 can receive various instructions input by an operator and send the instructions to the processor 201 for execution. The input device 204 can include a keyboard or a pointing device (e.g., a mouse, a trackball, a touchpad, or a touchscreen, etc.).

[0099] The display device 206 can display the results obtained by the processor 201 executing the instructions.

[0100] The memory 202 can store programs and data necessary for the operation of the operating system 2021, as well as intermediate results and other data during the computation of the processor 201.

[0101] It is to be understood that the memory 202 in the embodiments of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. The non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM) used as an external cache. The memory 202 of the apparatus and method described herein is intended to include, but is not limited to, these and any other suitable type of memory 202.

[0102] In some embodiments, the memory 202 stores the following elements, executable modules or data structures, or a subset thereof, or an extended set thereof: an operating system 2021 and an application program 2022.

[0103] The operating system 2021 includes various system programs, such as a framework layer, a core library layer, a driver layer, and the like, for implementing various basic services and processing hardware-based tasks. The application programs 2022 include various application programs 2022, such as a browser, and the like, for implementing various application services. Programs implementing the method of the embodiments of the present application can be included in the application programs 2022.

[0104] The processor 201 described above, when invoking and executing the application programs 2022 and data stored in the memory 202, specifically, the programs or instructions stored in the application programs 2022, executes the steps of the public place thief detection method according to the above-described embodiments.

[0105] The method disclosed in the above-described embodiments of the present application can be applied to the processor 201 or implemented by the processor 201. The processor 201 can be an integrated circuit chip having a signal processing capability. In the implementation process, each step of the above-described method can be completed by an integrated logic circuit or an instruction in the form of software in the processor 201. The processor 201 described above can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, and can implement or execute the various methods, steps and logical block diagrams disclosed in the embodiments of the present application. The general-purpose processor can be a microprocessor or the processor 201 can also be any conventional processor 201 and the like. The steps of the method disclosed in conjunction with the embodiments of the present application can be directly embodied as a hardware coding processor for execution, or a combination of a hardware coding processor and software modules in the coding processor. The software modules can be located in a storage medium such as random access memory (RAM), a flash memory, a read-only memory (ROM), a programmable read-only memory (PROM), an electrically programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a register, a removable disk, or a storage medium in other forms in the art. The storage medium is located in the memory 202, and the processor 201 reads information in the memory 202 and combines it with hardware to complete the steps of the above-described method.

[0106] It can be understood that the embodiments described herein can be realized by hardware, software, firmware, middleware, microcode or a combination thereof. For hardware implementation, the processing unit can be implemented in one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), general-purpose processors, controllers, micro-controllers, microprocessors, other electronic units designed to perform the functions described in the present application, or a combination thereof.

[0107] For software implementation, the techniques herein can be implemented by means of a module for performing the function, such as procedures, functions, and so on. Software codes can be stored in the memory 202 and executed by the processor 201. The memory 202 can be implemented in the processor 201 or outside the processor 201.

[0108] In particular, the processor 201 is further configured to read a computer program, and execute the following steps: predicting and outputting an answer to a question asked by a user for the pile charging method.

[0109] The fourth aspect of the embodiments of the present application further provides a computer readable storage medium, which stores a computer program. The computer program is run by the processor 201, so that the processor 201 executes the steps of the public place thief detection method of the above-mentioned embodiments.

[0110] In several embodiments provided in the present application, it should be understood that the disclosed methods and devices can be implemented in other manners. For example, the above-mentioned device embodiments are merely schematic. For example, the division of the units is only a logical function division. There can be another division manner for the actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0111] In addition, each function unit in the embodiments of the present application can be integrated into one processing unit, or each unit can be a separate physical unit, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware, or in the form of hardware plus software function units.

[0112] The integrated unit implemented in the form of software function units can be stored in a computer readable storage medium. The above-mentioned software function unit stored in a storage medium includes a plurality of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute part of the steps of the transceiving method of the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0113] While certain specific embodiments of the application have been described in detail herein for the purposes of exemplification and to provide a thorough and enabling disclosure, it will be understood that the application is not limited to the particular embodiments described. Any modifications of the methods and materials described herein, which come within the scope and spirit of the application, are to be considered within the scope of the application. The scope of the application is to be determined by the claims appended hereto, which are to be construed in accordance with the principles of patent law.

Claims

1. A method for detecting thieves in public places, characterized in that, include: Humans are detected using a human key point detection model, and a human bounding box is calculated using the human key points. Images of the head, upper body, and lower body are cropped using the location information of the key points on the human body. The intersection-over-union (IoU) ratio between the human body frames is calculated, and images of the area between the thigh and neck are cropped using the positional information of key points on the human hands. The step of calculating the IoU ratio includes: storing the current frame, the Nth frame before the current frame, and the Nth frame after the current frame; calculating the IoU ratio of the Nth frame after the Nth frame; the formula for calculating the IoU ratio between the human body frames is: ， Where IoU represents the intersection-union ratio, and A and B represent the two human bounding boxes respectively; The relationship between consecutive frames is expressed as follows: ， Wherein, when the IoU of the first Nth frame t’ The IoU of the current frame is equal to 0. t Greater than 0, IoU in the last N frames t’’ When the value equals 0, theft is considered to have occurred. The cropped images of the area between the thighs and necks are input into the Cls-UNet model to obtain segmentation and classification results. If both the segmentation and classification results meet preset conditions, and the relationship between consecutive frames also meets preset conditions, a theft is determined to have occurred. The Cls-UNet model's segmentation categories include: hand, backpack, shoulder bag, mobile phone, and wallet, with a classification number of 2, where 0 indicates no theft and 1 indicates theft. The Cls-UNet model's judgment logic is that if the segmentation reaches the hand, and also falls into any one of the categories of backpack, shoulder bag, mobile phone, and wallet, and the Cls-UNet model's classification category channel output is 1, a theft is determined to have occurred. A classification model is used to determine whether a hat is worn, and the colors of clothes and pants are analyzed using the HSV color space. Based on the judgment results of the classification model and the analysis results of the HSV color space, an alarm sound is emitted to alert the person who was robbed.

2. The method for detecting thieves in public places according to claim 1, characterized in that, Also includes: Storing video footage of the theft as evidence for further investigation.

3. A public place thief detection system, applied to the public place thief detection method according to any one of claims 1-2, characterized in that, The detection system includes: The detection module detects people using a human key point detection model. The first calculation module calculates the human body frame based on the human body key points; The second calculation module is used to calculate the intersection-union ratio between the human body frames; The cropping module uses the positional information of key points on the human hand to crop out an image of the area between the thigh and neck. The logic analysis module is used to input the cropped images of the area between the thighs and necks of the two individuals into the Cls-UNet model to obtain segmentation and classification results. If both the segmentation and classification results meet preset conditions, and the relationship between consecutive frames meets preset conditions, it is determined that theft has occurred.

4. An electronic device, characterized in that, include: A processor and a memory, wherein computer program instructions are stored in the memory, wherein when the computer program instructions are executed by the processor, the processor causes the processor to perform the steps of the public place thief detection method according to any one of claims 1-2.

5. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to perform the steps of the public place thief detection method according to any one of claims 1-2.

Images