Property loss tracing method and device, electronic equipment and storage medium

By obtaining and analyzing the data in the campus information database, determining the priority and monitoring scope of property protection, and conducting facial comparison and behavioral video analysis, it solves the problem that stolen personnel are difficult to track after property is lost on campus, and realizes efficient property tracing .

CN118968586APending Publication Date: 2024-11-15SOYO TECH DEV CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410966650.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-18
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

On campus, it is difficult to detect people who steal items in a timely manner after the student’s property is lost, and it is difficult to efficiently trace the source of property loss.

Method used

By obtaining target location data information in the information database, including student number, seat number, face image and property data, determine the priority and time period of property protection, set the behavior monitoring range, and compare face images during the monitoring time period. , and determine the probability of stealing based on the behavioral video and trace the source.

Benefits of technology

It realizes efficient property tracing after students' property is lost, quickly finds stolen personnel, and improves traceability efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118968586A_ABST
    Figure CN118968586A_ABST
Patent Text Reader

Abstract

The embodiment of the invention discloses a property loss tracing method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining the data information of a target position in an information database, and enabling the data information to comprise a student number, a seat number, a face image, property data and course information; determining a property protection priority and a property protection time period of the target position based on the data information; determining a behavior monitoring range based on the data information and the property protection priority; in the property protection time period, comparing the face image of any target person appearing in the behavior monitoring range with the face image corresponding to the target position; if the comparison is not passed, caching a behavior video of the target person in the behavior monitoring range; based on the behavior video of the target person, determining a correlation probability of property loss of the target position caused by the target person; and under the condition that property loss exists at the target position, tracing is carried out according to the correlation probability of each target person.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to a method, device, electronic device and storage medium for tracing lost property. Background Art

[0002] On campus, students' belongings are usually kept in classrooms or dormitories, and sometimes property loss occurs due to theft by others. However, due to the large population in the school, students often cannot discover in time that their property is lost, and they cannot quickly identify the person who stole the property after it is lost.

[0003] Therefore, there is an urgent need for a method that can efficiently trace the source of lost property after the item is stolen. Summary of the invention

[0004] In order to solve the above-mentioned problems existing in the prior art, the embodiments of the present application provide a method, device, electronic device and storage medium for tracing lost property, which can efficiently trace the lost property of students after the property is lost.

[0005] In a first aspect, an embodiment of the present invention provides a method for tracing the source of property loss, including:

[0006] Acquire data information of a target location in an information database, wherein the data information includes a student ID, a seat number, a face image, property data, and course information;

[0007] Determine the property protection priority and property protection time period of the target location based on the data information;

[0008] Determining the scope of behavior monitoring based on the data information and the property protection priority;

[0009] During the property protection time period, the facial image of any target person appearing within the behavior monitoring range is compared with the facial image corresponding to the target location;

[0010] If the comparison fails, the behavior video of the target person within the behavior monitoring range is cached;

[0011] Based on the behavior video of the target person, determining the relative probability that the target person causes property loss at the target location;

[0012] In the event that property is lost at the target location, the source is traced based on the relevant probability of each target person.

[0013] In a second aspect, an embodiment of the present invention provides a property loss tracing device, including an acquisition unit and a processing unit;

[0014] The acquisition unit is used to acquire data information of a target location in an information database, wherein the data information includes a student number, a seat number, a face image, property data, and course information;

[0015] The processing unit is used to determine the property protection priority and property protection time period of the target location based on the data information;

[0016] Determining the scope of behavior monitoring based on the data information and the property protection priority;

[0017] During the property protection time period, the facial image of any target person appearing within the behavior monitoring range is compared with the facial image corresponding to the target location;

[0018] If the comparison fails, the behavior video of the target person within the behavior monitoring range is cached;

[0019] Based on the behavior video of the target person, determining the relative probability that the target person causes property loss at the target location;

[0020] In the event that property is lost at the target location, the source is traced based on the relevant probability of each target person.

[0021] In a third aspect, an embodiment of the present invention provides an electronic device, including:

[0022] A processor and a memory, wherein the processor is connected to the memory, the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the method as described in the first aspect.

[0023] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method as described in the first aspect.

[0024] In a fifth aspect, an embodiment of the present application provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer is operable to cause the computer to execute the method described in the first aspect.

[0025] Implementing the embodiments of the present application has the following beneficial effects:

[0026] In the implementation mode of the present application, the data information of the target location in the information database is first obtained, wherein the data information includes the student number, seat number, face image, property data and course information, then, the property protection priority and property protection time period of the target location are determined based on the data information, next, the behavior monitoring range is determined based on the data information and the property protection priority, then, within the property protection time period, the face image of any target person appearing within the behavior monitoring range is compared with the face image corresponding to the target location, further, if the comparison fails, the behavior video of the target person within the behavior monitoring range is cached, and finally, based on the behavior video of the target person, the relevant probability of the target person causing the property loss at the target location is determined, and in the case of property loss at the target location, the source is traced according to the relevant probability of each target person. Thus, the protection priority is determined by the importance of the property, the monitoring range is further determined, the face image recognition of the person appearing within the monitoring range is performed within the property protection time period, and the relevant probability of suspicious persons and property loss is determined according to the behavior video, so that after the student's property is lost, the property loss can be efficiently traced. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the background technology, the drawings required for use in the embodiments of the present invention or the background technology will be described below.

[0028] Figure 1 A flowchart of a method for tracing the source of lost property provided in an embodiment of the present application;

[0029] Figure 2 A flowchart of a method for determining property protection priority and property protection time period provided in an embodiment of the present application;

[0030] Figure 3 A flowchart of a method for calculating the correlation probability based on a behavior video of a target person provided in an embodiment of the present application;

[0031] Figure 4 A flowchart of a method for acquiring expression information and action information provided in an embodiment of the present application;

[0032] Figure 5 A flowchart of a method for calculating the correlation probability based on effective stay time, expression information and action information provided in an embodiment of the present application;

[0033] Figure 6 A functional module composition block diagram of a property loss tracing device provided in an embodiment of the present application;

[0034] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0036] The terms "first", "second", "third" and "fourth" etc. in the specification and claims of the present application and the drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or modules is not limited to the listed steps or modules, but optionally includes steps or modules that are not listed, or optionally includes other steps or modules inherent to these processes, methods, products or devices.

[0037] Reference to "embodiments" herein means that a particular feature, result, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0038] See also Figure 1 , Figure 1 A flowchart of a method for tracing the source of lost property provided in an embodiment of the present application. Figure 1 As shown, the method includes but is not limited to the following steps:

[0039] S101: Acquire data information of a target location in an information database.

[0040] Among them, the data information includes student ID, seat number, facial image, property data and course information.

[0041] Among them, property data may include important asset information of students, and may include students’ financial status, such as students’ income sources, family financial status, personal savings and other financial information. Property data may also include students’ consumption data, such as data on students’ consumption habits, consumption amounts, consumption areas, etc. Property data may also include students’ loan status, such as student loans, credit card overdrafts, etc. Property data may also include financial assistance received by students, such as scholarships and grants.

[0042] Specifically, the data information in the information database can be entered by the students, and the information entry device can be set on the student's seat. For example, the information entry device can be a simple display screen with multiple virtual function buttons set on the display screen, which can receive the student ID, seat number, face image, property data, course information and other data information entered by the student on the display screen. When the seat is changed, the information entered on the display screen can be received again, or the data information can be matched and adjusted according to the position of the seat change.

[0043] Specifically, the data information may also include student dormitory information. For entering dormitory information, dormitory information entry authority may be opened, and after students enter the dormitory data information, the data information may be stored in the information database.

[0044] Optionally, the information entry device can be an electronic form. For example, an electronic form can be provided at the seat to receive data information input by students through a touch screen or using a keyboard, mouse, etc.

[0045] Optionally, a QR code or barcode can be placed on the seat to receive data information scanned and entered by students using a mobile phone or a dedicated scanning device.

[0046] Optionally, information can be entered through radio frequency identification technology. For example, a radio frequency identification tag is assigned to each seat, and data information entered by students by reading the tag information with a radio frequency identification reader / writer can be received.

[0047] Optionally, the information input device can realize the three-dimensional input of the student's items and save the three-dimensional data of the student's items to the information database. The three-dimensional data can be obtained by receiving the data of the object modeling on the display screen by the student and obtaining the three-dimensional data according to the object modeling data, or by analyzing the three-dimensional data of the object through the object pictures uploaded by the student.

[0048] S102: Determine the property protection priority and property protection time period of the target location based on the data information.

[0049] The target location may be a student classroom, a student dormitory, or a student seat.

[0050] In one possible implementation, the property protection priority may be prioritized and sorted according to multiple preset amounts. For example, a first preset amount and a second preset amount are set, and the first preset amount is greater than the second preset amount. When the property equivalent amount of the property data of the target location in the data information exceeds the first preset amount, the property protection priority of the target location is determined to be the highest level. When the property equivalent amount of the property data of the target location in the data information is between the first preset amount and the second preset amount, the property protection priority of the target location is determined to be an intermediate level. When the property equivalent amount of the property data of the target location in the data information is lower than the second preset amount, the property protection priority of the target location is determined to be the lowest level. Of course, the property protection priority may also be classified more finely here to achieve better property protection for valuables.

[0051] In a possible implementation, the property protection priority of the target location may be determined based on the property data in the data information within the class.

[0052] Exemplarily, the property protection priority of the target location is determined based on the number of properties. Taking the number of properties owned by students in the class into consideration, students with more properties may need to be given a relatively higher protection priority. For example, some students bring more expensive learning equipment such as tablets, electronic dictionaries, professional drawing tools, etc. into the class. In terms of property protection, the protection priority for students with a large number of such properties will be higher. For example, student A brings a laptop, a tablet and a high-end e-reader for study, while student B only has basic textbooks and stationery. In this regard, student A's property protection priority is relatively higher.

[0053] For example, the property protection priority of the target location is determined based on the property value factor, and the priority is determined based on the value of the property owned by the students. Students who own high-value property should be given priority protection. For example, some students bring expensive professional musical instruments, limited edition model figures, etc. into the classroom for storage. For example, student C brings his professional-grade violin worth tens of thousands of yuan to school for practice and storage, then the property protection priority for student C will be higher.

[0054] For example, the property protection priority of the target location is determined based on the factors of property use and learning relevance. Based on the importance and relevance of students' property to learning activities, the property that is more important to learning and teaching activities will have a higher protection priority, such as a student's scientific research project model, important learning results for display, etc. For example, in a science class, student D's scientific invention model is a work that will soon participate in an important competition, and this work is very important for the course learning results display and competition, so the property protection priority of student D will be increased.

[0055] For example, the property protection priority of the target location is determined based on the property vulnerability factor. If the student's property is fragile, easy to damage, or easy to lose, then the protection priority of these students' properties also needs to be increased, such as students' glass crafts, small precision instruments, etc. For example, student E's homemade glass crafts are easily damaged during transportation and movement, so a higher property protection priority level is required for student E's property.

[0056] For example, the property protection priority of the target location is determined based on customized factors. The property protection priority can also be customized by students to determine which of their own properties need to be protected. If it is based on a special agreement between the school and the students, or some students have special needs, such as temporarily storing valuable property in the classroom due to participating in an event, or for some items with lower property value factors but higher significance to the target students, then the protection priority is determined based on these agreements and needs. For example, during the school's organized activities, student F is responsible for keeping the valuable props and equipment related to the activities temporarily stored in the classroom. During this period, the property protection priority of student F should be increased.

[0057] In a possible implementation, the above-mentioned factors for determining property protection priority are jointly considered, corresponding weights are determined for multiple factors according to their importance, and the property protection priority is comprehensively determined based on the weights, wherein the multiple factors are a combination of any multiple factors including property quantity factors, property value factors, property use and learning correlation factors, property vulnerability factors, and custom factors.

[0058] Specifically, see Figure 2 , Figure 2 A flowchart of a method for determining property protection priority and property protection time period provided in an embodiment of the present application. Figure 2 As shown, the method includes but is not limited to the following steps:

[0059] S201: Determine the property equivalent amount and property loss probability of the target location based on the property data.

[0060] Specifically, for cash, the property equivalent amount is the monetary value of the cash. For some property that changes due to market value, the property equivalent amount can be estimated based on its market value. The probability of property loss of the item can be determined based on the size and shape of the item in the property data, the frequency and scene of use of the item, environmental factors, etc. For example, small, light or special-shaped items may be more likely to be lost, items that are frequently used or used in specific scenarios may be more likely to be lost, and noisy, crowded or chaotic environments may increase the risk of item loss.

[0061] S202: Determine the property protection priority based on the property equivalent amount and the property loss probability.

[0062] Specifically, the property protection priority is divided into high priority, medium priority and low priority according to the equivalent amount of the property and the probability of property loss. The most important property with a higher equivalent amount or facing a higher risk of loss is determined as a high priority for property protection. These property areas may include key facilities on campus, storage areas for students' valuables, etc. In these areas, advanced monitoring equipment such as high-definition cameras can be installed. For properties with a certain importance and loss risk, they can be determined as medium priority for property protection. These property areas may include general office areas, public areas, storage areas for secondary equipment, etc. For properties with relatively low value and loss risk, they can be determined as low priority for property protection.

[0063] For example, large amounts of cash and items such as gold and silver jewelry need to be listed as high priorities for property protection due to their high property equivalent amount and high probability of property loss.

[0064] For example, if a student's property data shows that both the property equivalent amount and the property loss probability are 0, then the student may be determined to have a low priority for property protection, and the camera's behavioral monitoring range of the student's target location may be reduced accordingly.

[0065] S203: Determine the property protection time period of the target location according to the course information and the face image.

[0066] Specifically, the property protection time period of the classroom is determined according to the course information as the break time and the school dismissal time, or the time period for courses that require outdoor activities, such as physical education classes, which also require the protection of student classroom property information. The property protection time period of the dormitory is determined according to the course information as the student class time period.

[0067] Specifically, the target position is monitored by monitoring equipment. When it is detected that the student is not in the seat in the classroom, this time period can also be used as the property protection time period of the classroom.

[0068] Optionally, sensors and monitoring equipment can be used, for example, sensors such as pressure sensors, temperature sensors, etc., to detect the usage of seats. When the seats are not used by students, this time period is also used as the property protection time period of the classroom.

[0069] S103: Determine the scope of behavior monitoring based on data information and property protection priority.

[0070] Specifically, the behavior monitoring range is determined based on the property data and seat number in the data information, the property protection priority and the structural characteristics of the space where the property is placed. For example, for a classroom where the property is placed, the frequency of personnel flow at the target location can be determined based on the seat number and the spatial structural characteristics of the classroom. For example, the frequency of personnel flow at locations near doors, windows and corridors is relatively high. The frequency of personnel flow here can be determined based on the historical personnel flow data of the target location. Monitoring weights are further set for the property protection priority and the frequency of personnel flow. Combined with the monitoring weights, the higher the level of property protection priority, the greater the frequency of personnel flow at the target location, and the larger the behavior monitoring range. Furthermore, the behavior monitoring range is also determined based on the seat number information in the data information and the structural characteristics of the space where the property is placed. For example, for a seat against a wall, the center point of the corresponding behavior monitoring range is no longer the center of the seat, but is biased towards the non-wall side. With the center point as the center, the preset monitoring range corresponding to the property protection priority is adjusted accordingly, and the adjusted range is used as the behavior detection range.

[0071] S104: During the property protection time period, the facial image of any target person appearing within the behavior monitoring range is compared with the facial image corresponding to the target position.

[0072] Specifically, a monitoring area is set up to ensure that the surveillance camera can cover the monitoring area during the property protection period, and can clearly capture the facial image of any target person appearing within the monitoring range, and compare the captured facial image with the facial image of the student in that seat.

[0073] S105: If the comparison fails, the behavior video of the target person within the behavior monitoring range is cached.

[0074] Specifically, ensure that the monitoring system is able to record and store videos, and cache the behavior videos of the target person within the behavior monitoring range when the comparison results are inconsistent.

[0075] S106: Based on the behavior video of the target person, determine the probability that the target person causes property loss at the target location.

[0076] Specifically, see Figure 3 , Figure 3 The flowchart of a method for calculating the correlation probability based on the behavior video of a target person provided in an embodiment of the present application is as follows. Figure 3 As shown, the method includes but is not limited to the following steps:

[0077] S301: Obtain the effective stay time of the target person at the target location in the behavior video.

[0078] Specifically, the time of the video clip in which the target person effectively stays at the target location in each behavior video is determined using the timestamp function of the monitoring system. Use the viewing tools or software provided by the monitoring system to identify the video records in chronological order and filter out the clips where the target person appears. For the clips where the target person appears, record the start time and end time of the clip. Subtract the start time from the end time to get the duration of the target person's appearance in the monitoring area. If multiple time periods or multiple target persons need to be analyzed, the duration of appearance in each time period can be summarized and counted to get the effective stay time.

[0079] S302: Sort the target personnel based on the effective stay time to obtain the target personnel sequence number.

[0080] Specifically, the longer the target person stays, the greater the probability that he or she is associated with property loss. The target persons are sorted in order of the length of their effective stay time to obtain the target person serial number.

[0081] S303: Acquire the expression information and action information of the target person based on the target person's serial number.

[0082] Specifically, see Figure 4 , Figure 4 The following is a flow chart of a method for acquiring expression information and action information provided in an embodiment of the present application. Figure 4 As shown, the method includes but is not limited to the following steps:

[0083] S401: Analyze the behavior video frame by frame according to the target person's serial number.

[0084] Specifically, the facial expression changes, body language, background and situation in the behavior video of each target person are analyzed frame by frame according to the target person's serial number. For example, the facial expression of the target person is monitored, and subtle changes in facial expression, such as eye contact, the arc of the corners of the mouth, the frown of the eyebrows, etc., the posture, gestures and body movements of the person are monitored, and frame by frame analysis is performed in combination with the background environment of the video or picture and the situation in which the person is located.

[0085] Furthermore, based on adjacent frames, the differences between frames are compared, and the image changes between multiple different frames are compared to determine the evolution process of expressions and actions.

[0086] Furthermore, in order to improve the analysis efficiency, key frames in the behavior video can be selected for analysis.

[0087] S402: Filter out pictures with preset expressions or preset actions as key frames.

[0088] Specifically, the preset actions include: observing the environment, hiding items, diverting attention, moving quickly, disguising behavior, avoiding surveillance, covering up traces, etc. For example, looking around to see if there are surveillance cameras, security guards or other personnel around, hiding stolen items in clothes, bags or other items, creating interference or distracting others, quickly picking up items and leaving the scene, avoiding the shooting range of surveillance cameras, or destroying surveillance equipment, trying to cover up traces of theft, such as restoring the original location of the items or cleaning up the scene, etc.

[0089] Optionally, pictures with preset expressions or preset actions can be screened out through video analysis software and expression recognition software. For example, specialized video analysis software can be used to detect and track the actions of the target person, such as opening a door, picking up an object, looking around, and other preset actions.

[0090] Optionally, pictures with preset expressions or preset actions are screened out through target tracking technology and facial expression analysis technology. For example, a target tracking algorithm is used to follow the movement of a target person in a video to obtain information about his preset actions.

[0091] Optionally, pictures with preset expressions or preset actions can be screened out through machine learning, deep learning and expression indicators. For example, by training a machine learning model or using deep learning technology, the target person in the behavior video recording can be classified and identified, and the information of preset actions can be extracted. The expression of the target person can be evaluated based on some common expression indicators to capture the preset expressions of the target person, such as smiling, frowning, surprise, etc.

[0092] S403: Acquire the expression information and action information of the target person according to the key frame.

[0093] Optionally, specialized expression recognition software is used to obtain the expression information and action information of the target person based on key frames. These software can analyze facial expressions in images or videos and provide relevant expression information.

[0094] Optionally, the expression information and action information of the target person are obtained according to the key frames through motion detection algorithms and sentiment analysis algorithms. For example, some software or programming tools can apply motion detection algorithms to identify the motion area in the video, and further analyze the action of the target person according to the key frames to obtain the action information. For example, sentiment analysis algorithms can analyze the sentiment information in the video or picture, including the expression information of the target person.

[0095] Optionally, some advanced computer vision and artificial intelligence technologies can be used for facial expression analysis, for example, by detecting and identifying facial features of the target person in key frames and classifying the expressions of the target person, so as to obtain the expression information of the target person.

[0096] S304: Calculate the relevant probability based on the effective stay time, expression information and action information.

[0097] Specifically, see Figure 5 , Figure 5 A flowchart of a method for calculating the correlation probability based on effective stay time, expression information and action information provided in an embodiment of the present application. Figure 5 As shown, the method includes but is not limited to the following steps:

[0098] S501: Calculate a first correlation probability according to the effective stay time.

[0099] Specifically, the time from when the target person enters the boundary of the monitoring range corresponding to the target position to when he leaves the boundary of the range is taken as the effective stay time.

[0100] Optionally, the first relevant probability is calculated according to the effective stay time and the total time of the behavior video, wherein the first relevant probability is the effective stay time divided by the total time of the behavior video multiplied by 100 percent.

[0101] Optionally, the first relevant probability is calculated according to the effective stay time and the average time of the behavior video, wherein the first relevant probability is the effective stay time divided by the average time of the behavior video multiplied by 100 percent.

[0102] S502: Calculate a second correlation probability according to the expression information.

[0103] Specifically, whether there is an expression stealing intention is analyzed based on the expression information, and the second relevant probability is calculated based on the number of occurrences or frequency of the expression stealing intention in the behavior video. An expression feature library is established to identify expression features that may be related to theft intention. These expression features can be obtained by analyzing and summarizing the expression performance in a large number of known theft cases. For example, nervous eyes, frequent gaze shifts, unnatural twitching of facial muscles, subtle changes in the corners of the mouth, etc. may be regarded as expression features related to theft intention. After obtaining the expression data in the behavior video, the expression data is processed and analyzed using image recognition and expression analysis technology to determine whether there are pre-set expression features related to theft intention. The second relevant probability is calculated based on the number of occurrences or frequency of the expression stealing intention. Assuming that the behavior of the target person is continuously identified and recorded within a preset period of time, if 30 out of 100 identifications result in the identification of expressions that are identified as having theft intention, then the second relevant probability is 0.3. Alternatively, the second relevant probability increases by 10% for each occurrence of expression stealing intention.

[0104] S503: Calculate a third correlation probability according to the action information.

[0105] Specifically, whether there is an action stealing intention is analyzed based on the action information, and the third related probability is calculated based on the number of occurrences or frequency of the action stealing intention in the behavior video. Before actually calculating the third related probability, a calculation analysis model is established and trained. First, a large amount of image or video data containing various normal actions and possible stealing-related actions is collected as a training data set. Then, through manual annotation or other methods, the actions in these data are marked as having stealing intentions. Next, the training data can be learned using models such as convolutional neural networks in deep learning. The model can automatically extract features from the action image and learn the association between these features and stealing intentions. In actual calculation, the action information in the behavior video is input into the trained model, and the model outputs a prediction on whether there is a stealing intention. A time window is set, and the number of occurrences of the action stealing intention in the time window is counted. Regarding the calculation of the third related probability based on the number of occurrences or frequency of the action stealing intention in the behavior video, the calculation of the third related probability can be expressed as the third related probability being the number of occurrences of the action stealing intention divided by the total number of recognitions. For example, in 100 recognitions, 20 occurrences of the action stealing intention are recognized, and then the third related probability is 0.2. Alternatively, each time an action stealing intention occurs, the third related probability increases by 15%.

[0106] In a possible implementation, a large amount of image and video data containing various possible theft actions and normal behaviors is collected, and the collected data is annotated to mark the parts of the image or video clips where the theft actions exist, as well as the specific types and characteristics of these actions. Then, the input behavior video is preprocessed, including operations such as cropping, scaling, graying, denoising, and normalization to reduce noise and irrelevant information in the behavior video and improve the accuracy and efficiency of subsequent processing. Human posture estimation technology is used to detect the joint point position and body contour of the target person in the behavior video. By analyzing the position and motion trajectory of the joint points, the action posture and movement direction of the person can be preliminarily determined. Then, based on the results of human posture estimation, features related to theft actions are extracted, such as arm extension, grasping action, body bending, action of approaching objects, displacement of objects, etc. Here, the optical flow method can be used to detect the movement direction and speed of pixels to analyze the action changes of the person; features such as shape, contour, texture, etc. can also be extracted to describe the form of the action. Next, machine learning or deep learning models are used to classify and identify the extracted action features to identify whether there is a theft action. For example, for a person stealing an item from a desk, first determine the position and posture of the person's arms and body through human posture estimation, then extract the action features of the arm reaching out to the item, grabbing the item, and putting the item into a pocket or other hidden location through optical flow or other feature extraction methods, and input these features into the trained classification model to determine whether the action is a theft action. Furthermore, the accuracy of recognition can be improved by combining scene information, such as the location of the item, the distance between the person and the item, and the contact situation.

[0107] S504: Calculate the correlation probability based on the first correlation probability, the second correlation probability and the third correlation probability.

[0108] Optionally, the relevant probability is calculated according to the probability obtained by adding the first relevant probability, the second relevant probability and the third relevant probability.

[0109] Optionally, the relevant probability is calculated according to the probability obtained by multiplying the first relevant probability, the second relevant probability and the third relevant probability.

[0110] Optionally, the first weight, the second weight and the third weight are determined for the first relevant probability, the second relevant probability and the third relevant probability respectively, and the relevant probability is the sum of the first weight multiplied by the first relevant probability, the second weight multiplied by the second relevant probability and the third weight multiplied by the third relevant probability. The method for determining the weight is determined by the historical theft data corresponding to the scenario. For each case in the historical theft case data, the values ​​of the first relevant probability, the second relevant probability and the third relevant probability calculated at that time are recorded, such as the mean, median, mode and other statistics of the first relevant probability, the second relevant probability and the third relevant probability, and whether the theft behavior actually occurred in the case. Then calculate the correlation coefficient between each relevant probability and the theft result. The correlation coefficient can measure the linear correlation degree between the relevant probability and the occurrence of theft, such as the Pearson correlation coefficient. Then, take the theft result as the dependent variable and the three relevant probabilities as the independent variables, and establish a regression equation. The regression coefficient in the regression equation can reflect the influence of each relevant probability on the theft result to a certain extent. If the mean, median or mode of a certain correlation probability is higher in the case of theft, then a higher weight can be given to the correlation probability. If the correlation coefficient between a certain correlation probability and the theft result is higher, it means that the correlation probability is more important in judging the occurrence of theft and is given a higher weight. In the regression analysis, the correlation probability with a larger regression coefficient has a greater impact on the theft result and is given a higher weight accordingly. For example, suppose that after analysis, it is found that in the historical data, the correlation coefficient between the first correlation probability and the theft result is 0.4, the correlation coefficient of the second correlation probability is 0.2, and the correlation coefficient of the third correlation probability is 0.4. Then it can be preliminarily determined that the weight of the first correlation probability is 0.4, the weight of the second correlation probability is 0.2, and the weight of the third correlation probability is 0.4.

[0111] S107: In the event of property loss at the target location, traceability is performed based on the relevant probability of each target person.

[0112] Specifically, when there is property loss at the target location, a tracing request for the target location is received from the client; in response to the tracing request, multiple target persons are sorted in order from high to low based on the relevant probability of each target person at the target location to obtain a sorting result. Then, the behavior video of each target person in the sorting result is obtained from the cache in turn; then, the behavior video of each target person is returned to the client in turn, so that the behavior video of each target person is played in turn on the client. It can be seen that in order of relevant probability from high to low, the behavior video of the target person with a higher relevant probability can be returned first, so that the person who caused the property loss can be quickly traced back and the tracing efficiency can be improved. Among them, the client can be a computer or a background server.

[0113] First of all, it is explained that when a user browses the behavior video of each target person on the client, if the source of the property loss is found, the client can send a traceability completion instruction, which is used to indicate that the traceability is completed to stop the traceability behavior for the target location. In one embodiment of the present application, in the process of returning the behavior video of each target person to the client in sequence, the sending time of the behavior video of each target person can be determined first. If the traceability completion instruction is not received at the sending time of the behavior video of each target person, the behavior video of each target person is returned to the client at the sending time.

[0114] Exemplarily, for the ith target person, after the behavior video of the i-1th target person is returned to the client, the sending time of the behavior video of the ith target person is determined based on the total playback time of the behavior video of the i-1th target person, the memory of the behavior video of the ith target person, and the network conditions. The ith target person is any target person. Specifically, based on the memory of the behavior video of the i-th target person and the network status, the total transmission time of the behavior video of the i-th target person is determined; based on the total playback time of the behavior video of the i-1-th target person and the total transmission time of the behavior video of the i-th target person, the playable time of the behavior video of the i-1-th target person is determined, wherein the playable time is the difference between the total playback time and the total time, and within the playable time, only the behavior video of the i-1-th target person is played, and there is no need to transmit the behavior video of the i-th target person; finally, based on the transmission completion time and the playable time of the behavior video of the i-1-th target person, the sending time of the behavior video of the i-th target person is determined, wherein the sending time is the time corresponding to the sum of the transmission completion time and the playable time of the behavior video of the i-1-th target person, and the transmission completion time can also be understood as the time to start playing the i-1-th target person.

[0115] It should be noted that if a tracing completion instruction is received before the sending time of the behavior video of a target person, the sending time of the behavior videos of the remaining target persons after the target person will no longer be calculated, and there is no need to send the behavior videos of the remaining target persons.

[0116] It should be noted that, since not every target person will cause property loss, some target person's behavior videos do not need to be sent, and in the process of tracing the source, it is very likely that the tracing has been completed based on a certain behavior video. Therefore, in this embodiment, each video will not be sent directly to the client to avoid sending too many useless videos. In addition, in order not to delay the tracing process, this application will calculate the sending time of each target person's behavior video and leave a certain amount of time for the transmission of each target person's behavior video, so that after completing the tracing of the previous target person, the next target person's behavior video will just be transmitted, which will not delay the tracing progress, and will not transmit too many videos.

[0117] Accordingly, after property is lost, multiple target persons are sorted according to relevant probabilities. Students can watch the behavioral video recordings in the order of sorting, and can quickly find the person most related to the property loss, thereby improving the efficiency of tracing the source after students' property is lost.

[0118] In a possible implementation, after the target thief is determined, the time of property loss can be determined accordingly. After the property loss time, the behavior video data related to the target thief is read from multiple related behavior videos, and the loss path of the lost item is further determined based on the behavior video data. For example, after the target thief is determined, by reading the behavior video data related to the person after the property loss time, if there is loss path reference data such as surveillance video records, access control system records, and electronic equipment positioning or use records, the target person's action trajectory after the theft is restored based on the loss path reference data, thereby inferring the loss path of the lost item. For example, if a book is lost in a school library, a student is determined to be suspected of theft through monitoring, and the approximate time of the book loss is known, then the relevant surveillance cameras in the library and the surrounding campus can be checked to record the student's action route after this time, or if valuables are lost in an office, the target thief's action trajectory can be tracked through access control system records, monitoring in the office area, and monitoring of roads around the company, and other behavior video data, and then the possible transfer path of the item can be inferred, further improving the efficiency of tracing the source after the student's property is lost.

[0119] In a possible implementation, the personnel information of each dormitory is first entered, and the student information for each seat in the classroom is also pre-entered, that is, the seats are numbered. Among them, the seats can be numbered by setting an information entry device on the seat, such as a simple display screen, and students can enter their own information on the display screen to achieve seat numbering. When students change seats, they can also re-enter information on the display screen. Then, in the target time period, for example, for the dormitory, the time period when students go to class can be used as the target time period to obtain whether there are target personnel going there, that is, to determine whether the personnel going to the target location are consistent with the information of the entered personnel. If not, it is considered that there is a risk of property theft, and this behavior will be videotaped and recorded for subsequent backtracking, and can be prompted by voice. For the classroom, the target time period can be the time period after school. If it is identified that someone other than the person goes to his seat within the target time period, the corresponding personnel information is recorded for subsequent backtracking, and voice reminders are performed, etc.

[0120] In summary, in the implementation mode of the present application, first obtain the data information of the target location in the information database, wherein the data information includes the student number, seat number, facial image, property data and course information, then determine the property protection priority and property protection time period of the target location based on the data information, next, determine the behavior monitoring range based on the data information and the property protection priority, then, within the property protection time period, compare the facial image of any target person appearing within the behavior monitoring range with the facial image corresponding to the target location, further, if the comparison fails, cache the behavior video of the target person within the behavior monitoring range, and finally, based on the behavior video of the target person, determine the relevant probability that the target person causes the property loss at the target location, and in the case of property loss at the target location, trace the source according to the relevant probability of each target person. Thus, the protection priority is determined by the importance of the property, the behavior monitoring range is further determined, the facial image recognition of the person appearing within the behavior monitoring range is performed within the property protection time period, and the relevant probability of suspicious persons and property loss is determined according to the behavior video, and then after the student's property is lost, the property loss can be efficiently traced according to the relevant probability.

[0121] See also Figure 6 , Figure 6 This is a functional module block diagram of a property loss tracing device provided in an embodiment of the present application. The property loss tracing device 600 includes:

[0122] The acquisition module 601 is used to acquire data information of a target location in an information database, wherein the data information includes a student number, a seat number, a face image, property data and course information;

[0123] A processing module 602 is used to determine the property protection priority and property protection time period of the target location based on the data information;

[0124] Determine the scope of behavioral monitoring based on data information and property protection priorities;

[0125] During the property protection period, the facial image of any target person appearing within the behavior monitoring range is compared with the facial image corresponding to the target location;

[0126] If the comparison fails, the behavior video of the target person within the behavior monitoring range is cached;

[0127] Based on the target person's behavior video, determine the relative probability that the target person will cause property loss at the target location;

[0128] In the event of property loss at the target location, traceability is performed based on the relative probability of each target person.

[0129] In some possible implementations, in determining the probability of the target person causing property loss at the target location based on the behavior video of the target person, the processing module 602 is specifically configured to:

[0130] Obtain the effective stay time of the target person at the target location in the behavior video;

[0131] Sort the target personnel based on their effective stay time to obtain their sequence numbers;

[0132] Acquire the target person's expression information and action information based on the target person's serial number;

[0133] The relevant probability is calculated based on the effective stay time, expression information and action information.

[0134] In some possible implementations, in terms of calculating the relevant probability based on the effective stay time, the expression information, and the action information, the processing module 602 is specifically configured to:

[0135] Calculate the first relevant probability according to the effective stay time;

[0136] Calculating a second related probability according to the expression information;

[0137] Calculating a third related probability based on the action information;

[0138] The correlation probability is calculated based on the first correlation probability, the second correlation probability, and the third correlation probability.

[0139] In some possible implementations, in calculating the first relevant probability according to the effective stay time, the processing module 602 is specifically configured to:

[0140] Calculate a first correlation probability according to the effective stay time and the total time of the behavior video, wherein the first correlation probability is the effective stay time divided by the total time of the behavior video multiplied by 100%;

[0141] Alternatively, the first correlation probability is calculated based on the effective dwell time and the average time of the behavior video.

[0142] In some possible implementations, in calculating the correlation probability based on the first correlation probability, the second correlation probability, and the third correlation probability, the processing module 602 is specifically configured to:

[0143] Calculating the relevant probability according to the probability obtained by adding the first relevant probability, the second relevant probability and the third relevant probability;

[0144] Alternatively, the relevant probability is calculated according to the probability obtained by multiplying the first relevant probability, the second relevant probability and the third relevant probability.

[0145] In some possible implementations, in terms of acquiring the expression information and action information of the target person based on the target person serial number, the processing module 602 is specifically configured to:

[0146] Analyze the behavior video frame by frame according to the target person's serial number;

[0147] Filter out pictures with preset expressions or preset actions as key frames;

[0148] Obtain the target person's expression and action information based on the key frames.

[0149] In some possible implementations, in determining the property protection priority and property protection time period of the target location based on the data information, the processing module 602 is specifically configured to:

[0150] Determine the property equivalent amount and the probability of property loss at the target location based on the property data;

[0151] Determine property protection priorities based on property equivalent amount and probability of property loss;

[0152] The property protection time period of the target location is determined based on the course information and face image.

[0153] See also Figure 7 , Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 7As shown, the electronic device 700 includes a transceiver 701, a processor 702 and a memory 703. They are connected via a bus 704. The memory 703 is used to store computer programs and data, and can transmit the data stored in the memory 703 to the processor 702. The electronic device can be the above-mentioned property loss tracing device, the processor 702 can be the above-mentioned processing module 602, and the transceiver 701 can be the above-mentioned acquisition module 601.

[0154] The transceiver 701 is used to perform the following operations:

[0155] Obtain data information of a target location in an information database, wherein the data information includes student ID, seat number, face image, property data and course information.

[0156] The processor 702 is used to read the computer program in the memory 703 and perform the following operations:

[0157] Determine the property protection priority and property protection time period of the target location based on the data information;

[0158] Determine the scope of behavioral monitoring based on data information and property protection priorities;

[0159] During the property protection period, the facial image of any target person appearing within the behavior monitoring range is compared with the facial image corresponding to the target location;

[0160] If the comparison fails, the behavior video of the target person within the behavior monitoring range is cached;

[0161] Based on the target person's behavior video, determine the relative probability that the target person will cause property loss at the target location;

[0162] In the event of property loss at the target location, traceability is performed based on the relative probability of each target person.

[0163] In some possible implementations, in determining the probability of the target person causing property loss at the target location based on the target person's behavior video, the processor 702 is specifically configured to perform the following operations:

[0164] Obtain the effective stay time of the target person at the target location in the behavior video;

[0165] Sort the target personnel based on their effective stay time to obtain their sequence numbers;

[0166] Acquire the target person's expression information and action information based on the target person's serial number;

[0167] The relevant probability is calculated based on the effective stay time, expression information and action information.

[0168] In some possible implementations, in terms of calculating the relevant probability based on the effective stay time, the expression information, and the action information, the processor 702 is specifically configured to perform the following operations:

[0169] Calculate the first relevant probability according to the effective stay time;

[0170] Calculating a second related probability according to the expression information;

[0171] Calculating a third related probability based on the action information;

[0172] The correlation probability is calculated based on the first correlation probability, the second correlation probability, and the third correlation probability.

[0173] In some possible implementations, in terms of calculating the first relevant probability according to the effective stay time, the processor 702 is specifically configured to perform the following operations:

[0174] Calculate a first correlation probability according to the effective stay time and the total time of the behavior video, wherein the first correlation probability is the effective stay time divided by the total time of the behavior video multiplied by 100%;

[0175] Alternatively, the first correlation probability is calculated based on the effective dwell time and the average time of the behavior video.

[0176] In some possible implementations, in terms of calculating the correlation probability based on the first correlation probability, the second correlation probability, and the third correlation probability, the processor 702 is specifically configured to perform the following operations:

[0177] Calculating the relevant probability according to the probability obtained by adding the first relevant probability, the second relevant probability and the third relevant probability;

[0178] Alternatively, the relevant probability is calculated according to the probability obtained by multiplying the first relevant probability, the second relevant probability and the third relevant probability.

[0179] In some possible implementations, in terms of acquiring the expression information and action information of the target person based on the target person serial number, the processor 702 is specifically configured to perform the following operations:

[0180] Analyze the behavior video frame by frame according to the target person's serial number;

[0181] Filter out pictures with preset expressions or preset actions as key frames;

[0182] Obtain the target person's expression and action information based on the key frames.

[0183] In some possible implementations, in determining the property protection priority and property protection time period of the target location based on the data information, the processor 702 is specifically configured to perform the following operations:

[0184] Determine the property equivalent amount and the probability of property loss at the target location based on the property data;

[0185] Determine property protection priorities based on property equivalent amount and probability of property loss;

[0186] The property protection time period of the target location is determined based on the course information and face image.

[0187] It should be understood that the property loss tracing device of the present application can be a server or other computing device, wherein the server can be a cloud computing server, a content delivery network (Content Delivery Network, CDN) server, a network time protocol (Network Time Protocol, NTP), a domain name resolution system (Domain Name System, DNS) server, and other various types of servers.

[0188] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. The computer program is executed by a processor to implement part or all of the steps of any method for tracing the source of property loss as recorded in the above method embodiments.

[0189] An embodiment of the present application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute part or all of the steps of any one of the methods for tracing the source of property loss recorded in the above method embodiments.

[0190] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present application.

[0191] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0192] In the several embodiments provided in the present application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of modules, which is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or modules can be electrical or other forms.

[0193] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0194] In addition, each functional module in each embodiment of the present application can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or software program modules.

[0195] If the integrated module is implemented in the form of a software program module and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a memory and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of each embodiment method of the present application. The aforementioned memory includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk and other media that can store program codes.

[0196] A person skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable memory, and the memory can include: a flash drive, a read-only memory (English: Read-Only Memory, abbreviated as: ROM), a random access memory (English: Random Access Memory, abbreviated as: RAM), a magnetic disk or an optical disk, etc.

[0197] The embodiments of the present application are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for tracing the source of property loss, characterized in that: include: Acquire data information of a target location in an information database, wherein the data information includes a student ID, a seat number, a face image, property data, and course information; Determine the property protection priority and property protection time period of the target location based on the data information; Determining a behavior monitoring scope based on the data information and the property protection priority; During the property protection time period, the facial image of any target person appearing within the behavior monitoring range is compared with the facial image corresponding to the target location; If the comparison fails, the behavior video of the target person within the behavior monitoring range is cached; Based on the behavior video of the target person, determining the relative probability that the target person causes property loss at the target location; In the event that property is lost at the target location, the source is traced based on the relevant probability of each target person.

2. The method according to claim 1, characterized in that The determining, based on the behavior video of the target person, the probability of the target person causing property loss at the target location includes: Obtaining the effective stay time of the target person at the target location in the behavior video; Sort the target personnel based on the effective stay time to obtain the target personnel sequence number; Acquire the expression information and action information of the target person based on the target person serial number; The relevant probability is calculated based on the effective stay time, the expression information and the action information.

3. The method according to claim 2, characterized in that The calculating the relevant probability based on the effective stay time, the expression information and the action information comprises: Calculate a first relevant probability according to the effective stay time; Calculating a second correlation probability according to the expression information; Calculating a third correlation probability according to the action information; The correlation probability is calculated based on the first correlation probability, the second correlation probability, and the third correlation probability.

4. The method according to claim 3, characterized in that The calculating the first relevant probability according to the effective stay time comprises: Calculate a first correlation probability according to the effective stay time and the total time of the behavior video, wherein the first correlation probability is the effective stay time divided by the total time of the behavior video multiplied by 100%; Alternatively, the first correlation probability is calculated according to the effective stay time and the average time of the behavior video.

5. The method according to claim 3 or 4, characterized in that The calculating the correlation probability based on the first correlation probability, the second correlation probability and the third correlation probability comprises: Calculating the correlation probability according to the probability obtained by adding the first correlation probability, the second correlation probability and the third correlation probability; Alternatively, the relevant probability is calculated according to the probability obtained by multiplying the first relevant probability, the second relevant probability and the third relevant probability.

6. The method according to any one of claims 2 to 5, characterized in that: The acquiring the expression information and action information of the target person based on the target person serial number includes: Analyze the behavior video frame by frame according to the target person's serial number; Filter out pictures with preset expressions or preset actions as key frames; The expression information and action information of the target person are acquired according to the key frame.

7. The method according to any one of claims 1 to 6, characterized in that: The determining of the property protection priority and property protection time period of the target location based on the data information includes: Determine the property equivalent amount and the property loss probability of the target location based on the property data; Determining the property protection priority based on the property equivalent amount and the property loss probability; The property protection time period of the target location is determined according to the course information and the facial image.

8. A property loss tracing device, characterized in that: The property loss tracing device includes an acquisition unit and a processing unit; The acquisition unit is used to acquire data information of a target location in an information database, wherein the data information includes a student number, a seat number, a face image, property data, and course information; The processing unit is used to determine the property protection priority and property protection time period of the target location based on the data information; Determining a behavior monitoring scope based on the data information and the property protection priority; During the property protection time period, the facial image of any target person appearing within the behavior monitoring range is compared with the facial image corresponding to the target location; If the comparison fails, the behavior video of the target person within the behavior monitoring range is cached; Based on the behavior video of the target person, determining the relative probability that the target person causes property loss at the target location; In the event that property is lost at the target location, the source is traced based on the relevant probability of each target person.

9. An electronic device, characterized in that: include: A processor and a memory, the processor is connected to the memory, the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method according to any one of claims 1 to 7.