A method and system for visual detection of examinees and invigilators in examination room scenes
By using the examination desk detection model and target tracking algorithm in the examination room monitoring system, combining Y-axis coordinate sorting and spatial position judgment, the accurate distinction between candidates and invigilators is achieved, the problem of indistinguishability in examination room monitoring is solved, and abnormal behavior is improved Accuracy and efficiency of analysis.
Patent Information
- Application Number
- CN202411113061.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-14
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-08-14
AI Technical Summary
In the examination room monitoring scenario, it is impossible to effectively distinguish between candidates and invigilators, which makes it difficult to conduct accurate behavioral analysis and abnormal behavior detection.
By obtaining the target video stream, intercepting the video frame data as input, using the test desk detection model and the target tracking algorithm, combining the candidate-test desk matching algorithm, the target boxes of the candidate and the invigilator are separated, and the Y-axis coordinate sorting and spatial position judgment are used to determine the target boxes of the candidate and the invigilator. The camera direction achieves accurate classification of pedestrian target boxes.
It effectively solves the problem of being unable to distinguish between candidates and invigilators in the examination room monitoring scenario, provides more sufficient data to support the abnormal behavior analysis of smart inspection inspections, and improves the accuracy and efficiency of monitoring.
Smart Images

Figure CN118644809B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of smart examination inspection technology, and in particular to a method and system for visual detection of examinees and invigilators in an examination room scenario. Background Art
[0002] Smart patrol is a form of examination supervision that combines advanced technologies such as artificial intelligence, the Internet of Things, and big data analysis. It aims to improve the fairness, security, and efficiency of examinations, and overcomes the limitations of traditional manual proctoring through automation and intelligent means. The system captures video and audio information in the examination room in real time by deploying high-definition surveillance cameras, sensors, and other equipment in the examination room. Using artificial intelligence algorithms, the system can automatically identify suspicious behaviors of candidates, such as cheating and whispering. At the same time, through big data analysis, the system can predict potential violations, thereby achieving comprehensive monitoring of the examination process. In addition, smart patrol also reduces dependence on manual proctoring by optimizing human resource allocation. It not only improves the response speed and processing efficiency of proctoring, but also reduces the cost of supervision, providing an innovative supervision solution for education and examination departments. With the continuous advancement of technology, smart patrol is expected to play a more important role in future education examinations.
[0003] Smart inspection usually uses surveillance video as a data source and applies image algorithm technology for in-depth analysis. The system not only monitors candidates in real time, but also performs specific abnormal behavior analysis on the invigilator's behavior, thereby ensuring the fairness and transparency of the entire examination process. Through this meticulous monitoring, smart inspection can promptly detect and deal with violations in the examination room and maintain the seriousness and authority of the examination. However, since surveillance video is composed of a two-dimensional image frame sequence and does not have three-dimensional spatial information, it is difficult to classify disordered pedestrian target frames and distinguish candidates from invigilators after using algorithms such as target detection for pedestrian detection.
[0004] Therefore, there is an urgent need for a visual detection method for examinees and invigilators in examination room scenarios to effectively solve the technical problem of being unable to distinguish between examinees and invigilators in examination room monitoring scenarios. Summary of the invention
[0005] The present invention aims to provide a method and system for visually detecting examinees and invigilators in an examination room scenario, which can effectively solve the technical problem that examinees and invigilators cannot be distinguished in the examination room monitoring scenario.
[0006] In order to achieve the above object, the present invention provides the following technical solutions:
[0007] In a first aspect, an embodiment of the present invention provides a method for visually detecting examinees and invigilators in an examination room scenario, the method comprising the following steps:
[0008] Acquire a target video stream, wherein the frame data of the target video stream carries a storage timestamp of each frame data and channel identification information of the video stream to which each frame data belongs;
[0009] Intercept any frame of the target video stream as the input of the examination table detection model to obtain the target frames of all examination tables in the examination room;
[0010] Sort all the test table target frames according to the Y-axis order of the lower left corner coordinates to obtain the test table target frame with the longest distance on the Y-axis;
[0011] The Y-axis farthest distance test table target frame is calculated to determine the area position in the monitoring image to determine whether the current camera is the left camera or the right camera.
[0012] Optionally, intercepting any frame of the target video stream as the input of the examination table detection model to obtain all the examination table target frames in the examination room also includes:
[0013] Label all the target frames of the examination tables in the examination room to form a target detection frame set D;
[0014] The set D of object detection frames of the examination tables is sorted and numbered, and its ID is fixed and matched with the seat number of the examination room.
[0015] Optionally, sorting and numbering the test table object detection frame set D includes:
[0016] S1, traverse the test table target detection frame set D;
[0017] S2, searching for the target frame box_rt with the smallest distance from the X=0 axis in the target detection frame set D of the examination table, and determining the target frame as the first row and first column of the desk based on the prior rule;
[0018] S3, searching for the target frame box_ld with the largest distance from the Y=0 axis in the test desk target detection frame set D, and determining the target frame as the desk in the first row and the Nth column based on a priori rules;
[0019] S4, construct a straight line L based on the center points of the two boxes box_rt and box_ld, traverse the test table target detection box set D to filter out the target boxes that are penetrated by the straight line L, and form a set R,
[0020] The set R currently stores the target frames of the examination tables in the first row of the examination room;
[0021] S5, sorting the set R according to the size of the X-axis, and giving each frame a number in sequence;
[0022] S6, calculating the size of the set R to obtain the number of candidates in the row;
[0023] S7, removing all elements in the set R from the test table object detection frame set D;
[0024] S8, if the test table object detection frame set D is an empty set, sort and number all the test tables to form the set R;
[0025] If the examination table object detection frame set D is not an empty set, repeat the above steps until the examination table object detection frame set D is an empty set.
[0026] Optionally, sorting all the test table target frames according to the Y-axis order of the lower left corner coordinates further includes:
[0027] Pairing the elements in the set R that have been sorted and numbered in pairs along the Y axis in vertical order;
[0028] The candidate test area set E is formed by matching the diagonal lines of the target box;
[0029] After completing the examination table target detection task, execute the examination table target detection resource release operation.
[0030] Optionally, intercepting any frame of data of the target video stream also includes: inputting a pedestrian detection model and a target tracking algorithm into the target video stream frame by frame to obtain a set P of pedestrian target frames with tracking IDs; traversing the set P to calculate the center point P1 of each pedestrian target frame.
[0031] Optionally, the method further includes: determining whether the center point P1 is located within an element in the examinee's test area set E;
[0032] If the center point P1 is located in an element of the set E, then P1 corresponds to the candidate-associated frame number Sf+1 in the pedestrian attribute, and if Sf> the preset candidate threshold, the pedestrian is judged to be a candidate;
[0033] If the center point P1 is not in the set E, then P1 corresponds to the invigilator-associated frame number Tf+1 in the pedestrian attribute. If Tf>the preset invigilator threshold, the pedestrian is determined to be an invigilator.
[0034] Optionally, if the candidate has been identified as a candidate, the bound candidate test area is retrieved to determine whether the candidate detection frame is still within the area. If so, the verification is passed. Otherwise, the candidate pedestrian attribute information is destroyed and the relationship is re-established.
[0035] If the invigilator has been identified, traverse the candidate test area set E to determine whether the invigilator detection box is still in the area. If so, Tf-1. If Tf<the preset invigilator threshold, destroy the invigilator's pedestrian attribute information and re-establish the relationship; if the pedestrian detection box is not in any element area in the set E, the verification is passed.
[0036] In a second aspect, an embodiment of the present invention provides a visual detection system for examinees and invigilators in an examination room scenario, the system comprising:
[0037] An acquisition module is used to acquire a target video stream, wherein the frame data of the target video stream carries a storage timestamp of each frame data and channel identification information of the video stream to which each frame data belongs;
[0038] The model input module intercepts any frame data of the target video stream as the input of the examination table detection model to obtain the target frames of all examination tables in the examination room;
[0039] A sorting module sorts all the test table target frames according to the Y-axis order of the lower left corner coordinates to obtain the test table target frame with the longest distance on the Y-axis;
[0040] The judgment module calculates the area position of the farthest distance test table target frame on the Y axis in the monitoring image to determine whether the current camera is a left camera or a right camera.
[0041] In a third aspect, an embodiment of the present invention provides an electronic device, the electronic device comprising:
[0042] at least one processor;
[0043] at least one memory for storing at least one program;
[0044] When the at least one program is executed by the at least one processor, the at least one processor implements the visual detection method for examinees and invigilators for examination room scenarios as described in any one of the above items.
[0045] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, which stores a program executable by a processor, and the program executable by the processor is used to execute the visual detection method for examinees and invigilators for examination scenarios as described in any one of the above items when executed by the processor.
[0046] The beneficial effects of the present invention are as follows: the present invention discloses a method and system for visual detection of examinees and invigilators in an examination room scene. The present invention obtains a target video stream, wherein the frame data of the target video stream carries the storage timestamp of each frame data and the channel identification information of the video stream to which each frame data belongs; intercepts any frame data of the target video stream as the input of the examination table detection model to obtain all the examination table target frames in the examination room; sorts all the examination table target frames according to the Y-axis order of the lower left corner coordinate to obtain the examination table target frame with the farthest distance on the Y-axis; calculates the regional position of the examination table target frame with the farthest distance on the Y-axis in the monitoring image to determine whether the current camera is the left camera or the right camera. By obtaining the spatial position of the examination table as the relative reference system of the examination table target frame, combining the target tracking algorithm with the examinee-examination table matching algorithm to complete the separation of the examinee target frame in the pedestrian target frame, and then reversely separating the invigilator's target frame, the problem of being unable to distinguish between examinees and invigilators in the examination room monitoring scene is effectively solved, and more adequate preparation is provided for the analysis of abnormal behavior of intelligent inspection. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0048] Figure 1 A flowchart of a method for visually detecting examinees and invigilators in an examination room scenario provided in an embodiment of the present application;
[0049] Figure 2 A schematic diagram of the flow chart of the sorting and numbering algorithm provided in the embodiment of the present application;
[0050] Figure 3 A schematic diagram of the overall process of the method for visual detection of examinees and invigilators in an examination room scenario provided by an embodiment of the present application;
[0051] Figure 4 A schematic diagram of the overall process provided for the embodiments of the present application;
[0052] Figure 5 A schematic block diagram of a visual detection system for examinees and invigilators in an examination room scenario provided in an embodiment of the present application;
[0053] Figure 6 A schematic structural block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0054] The following will be combined with the embodiments and drawings to clearly and completely describe the concept, specific structure and technical effects of the present invention, so as to fully understand the purpose, scheme and effect of the present invention. It should be noted that the embodiments and features in the embodiments of the present invention can be combined with each other without conflict.
[0055] In the related technologies, in scenarios such as access control and attendance that require user identity authentication, most of them use access cards, facial recognition and other methods to authenticate identity. The access card method has the possibility of fraudulent use, while the facial recognition method often only collects the user's facial information, which has the problem of incomplete personal information.
[0056] To solve the problems in the background technology, the embodiments of the present invention provide the following technical solutions to improve the accuracy and comprehensiveness of identity authentication results.
[0057] See also Figure 1 , Figure 1 The present invention provides a method for visually detecting examinees and invigilators in an examination room, the method comprising the following steps:
[0058] S110, obtaining a target video stream, wherein the frame data of the target video stream carries a storage timestamp of each frame data and channel identification information of the video stream to which each frame data belongs;
[0059] S120, intercepting any frame of the target video stream as input of the examination table detection model to obtain target frames of all examination tables in the examination room;
[0060] S130, sorting all the test table target frames according to the Y-axis order of the lower left corner coordinates to obtain the test table target frame with the longest distance on the Y-axis;
[0061] S140, calculating the area position of the farthest distance test table target frame on the Y axis in the monitoring image to determine whether the current camera is a left camera or a right camera.
[0062] Exemplarily, a video processing library (such as OpenCV) is used to obtain a video stream, and a frame of image is captured from the video stream. The frame of image is input into the model to obtain all the target frames of the examination tables, and the target frames of the examination tables that are farthest from the Y axis (i.e., the bottom of the image) are obtained. The left or right area of the monitoring image where the frame is located is calculated to determine whether the current camera is the left camera or the right camera of the examination room.
[0063] In some embodiments of the present application, intercepting any frame of the target video stream as the input of the examination table detection model to obtain all the examination table target frames in the examination room also includes:
[0064] Label all the target frames of the examination tables in the examination room to form a target detection frame set D;
[0065] The set D of object detection frames of the examination tables is sorted and numbered, and its ID is fixed and matched with the seat number of the examination room.
[0066] For example, a unique ID is assigned to each detection box, and these IDs are kept unchanged in subsequent frames. The sorting criterion can be based on the Y-axis coordinate of the lower left corner of the detection box, and the numbering can start from 1. The ID is kept unchanged by using IOU (Intersection Over Union) tracking or using a tracker (such as KCF, CSRT, DeepSORT, etc.).
[0067] See also Figure 2 , Figure 2 A schematic diagram of the flow chart of the sorting and numbering algorithm provided by the present invention is shown in FIG. Figure 2 As shown, sorting and numbering the test table target detection frame set D includes:
[0068] S1, traverse the test table target detection frame set D;
[0069] S2, searching for the target frame box_rt with the smallest distance from the X=0 axis in the target detection frame set D of the examination table, and determining the target frame as the first row and first column of the desk based on the prior rule;
[0070] S3, searching for the target frame box_ld with the largest distance from the Y=0 axis in the test desk target detection frame set D, and determining the target frame as the desk in the first row and the Nth column based on a priori rules;
[0071] S4, construct a straight line L based on the center points of the two boxes box_rt and box_ld, traverse the test table target detection box set D to filter out the target boxes that are penetrated by the straight line L, and form a set R,
[0072] The set R currently stores the target frames of the examination tables in the first row of the examination room;
[0073] S5, sorting the set R according to the size of the X-axis, and giving each frame a number in sequence;
[0074] S6, calculating the size of the set R to obtain the number of candidates in the row;
[0075] S7, removing all elements in the set R from the test table object detection frame set D;
[0076] S8, if the test table object detection frame set D is an empty set, sort and number all the test tables to form the set R;
[0077] If the examination table object detection frame set D is not an empty set, repeat the above steps until the examination table object detection frame set D is an empty set.
[0078] Exemplarily, in S3, N is greater than or equal to 1.
[0079] For example, when the test room is a standard test room with 6 rows and 5 columns, box_rt is the desk in the first row and first column, and box_ld is the desk in the first row and fifth column.
[0080] After removing all elements in the set R from the examination table object detection frame set D in S7, the set D should now only have the examination table object frames from the 2nd row to the 6th row left.
[0081] In some embodiments of the present application, sorting all the test table target frames according to the Y-axis order of the lower left corner coordinates further includes: pairing the elements in the set R that have been sorted and numbered in a vertical order along the Y-axis;
[0082] The candidate test area set E is formed by matching the diagonal lines of the target box;
[0083] After completing the examination table target detection task, execute the examination table target detection resource release operation.
[0084] For example, when the test room is a standard test room with 6 rows and 5 columns, the diagonal lines of the paired target boxes constitute a set E of 30 (6x5) test area for candidates.
[0085] In some embodiments of the present application, intercepting any frame of data of the target video stream also includes: inputting a pedestrian detection model and a target tracking algorithm into the target video stream frame by frame to obtain a set P of pedestrian target frames with tracking IDs; traversing the set P to calculate the center point P1 of each pedestrian target frame.
[0086] It should be noted that OpenCV and deep learning models (such as YOLO or SSD) can be used for pedestrian detection, and combined with algorithms such as DeepSORT for target tracking. Traverse each pedestrian target box set P with a tracking ID and calculate the center point P1 of each pedestrian target box.
[0087] See also Figure 3 , Figure 3 A flow chart of a pedestrian detection model and a target tracking algorithm provided by the present invention; Figure 3 As shown, the method also includes:
[0088] S210, determining whether the center point P1 is located within an element in the candidate's test area set E;
[0089] S220, if the center point P1 is located in an element of the set E, then P1 corresponds to the number of test-taker associated frames Sf+1 in the pedestrian attribute, and if Sf> the preset test-taker threshold, the pedestrian is determined to be a test-taker;
[0090] S230, if the center point P1 is not in the set E, then P1 corresponds to the invigilator-associated frame number Tf+1 in the pedestrian attribute, and if Tf>the preset invigilator threshold, the pedestrian is determined to be an invigilator.
[0091] Please combine Figure 4 In some embodiments of the present application, if the candidate has been identified as a candidate, the bound candidate test area is taken out to determine whether the candidate detection frame is still in the area. If so, the verification is passed. Otherwise, the candidate pedestrian attribute information is destroyed and the relationship is re-established;
[0092] If the invigilator has been identified, traverse the candidate test area set E to determine whether the invigilator detection box is still in the area. If so, Tf-1. If Tf<the preset invigilator threshold, destroy the invigilator's pedestrian attribute information and re-establish the relationship; if the pedestrian detection box is not in any element area in the set E, the verification is passed.
[0093] For example, pedestrians are detected and tracked using DeepSORT. The center point P1 of each pedestrian target box is calculated. It is determined whether the center point is within the test area and the corresponding attribute Sf or Tf is updated. It is determined whether the pedestrian is a test taker or an invigilator based on a preset threshold.
[0094] It should be noted that pedestrian attributes include pedestrian IDs, and a corresponding information database can be established through pedestrian IDs; if a person has been identified as a candidate or an invigilator, it is checked whether he or she is still in the corresponding area, and his or her attribute information is updated or destroyed according to the conditions.
[0095] The embodiment of the present application provides a method for visual detection of examinees and invigilators in an examination room scene, by acquiring a target video stream, the frame data of which carries the storage timestamp of each frame data and the channel identification information of the video stream to which each frame data belongs; intercepting any frame data of the target video stream as the input of the examination table detection model to obtain all the examination table target frames in the examination room; sorting all the examination table target frames according to the Y-axis order of the lower left corner coordinate to obtain the examination table target frame with the farthest distance on the Y-axis; calculating the position of the area in the surveillance image where the examination table target frame with the farthest distance on the Y-axis is located to determine whether the current camera is the left camera or the right camera. By acquiring the spatial position of the examination table as the relative reference system of the examination table target frame, combining the target tracking algorithm with the examinee-examination table matching algorithm to complete the separation of the examinee target frame in the pedestrian target frame, and then reversely separating the invigilator's target frame, the problem of being unable to distinguish between examinees and invigilators in the examination room monitoring scene is effectively solved, and more adequate preparation is provided for the analysis of abnormal behavior of intelligent inspection.
[0096] See also Figure 5 The second aspect of the embodiment of the present application provides a visual detection system for examinees and invigilators in an examination room scenario, the system comprising:
[0097] The acquisition module 310 is used to acquire a target video stream, wherein the frame data of the target video stream carries a storage timestamp of each frame data and channel identification information of the video stream to which each frame data belongs;
[0098] The model input module 320 is used to intercept any frame data of the target video stream as the input of the examination table detection model to obtain the target frames of all examination tables in the examination room;
[0099] The sorting module 330 is used to sort all the test table target frames according to the Y-axis order of the lower left corner coordinates to obtain the test table target frame with the longest distance on the Y-axis;
[0100] The judgment module 340 is used to calculate the area where the farthest distance of the Y-axis test table target frame is located in the monitoring image to determine whether the current camera is a left camera or a right camera.
[0101] According to the above embodiments, the embodiments of the present application provide a method for visual detection of examinees and invigilators in an examination room scene, by acquiring a target video stream, the frame data of which carries the storage timestamp of each frame data and the channel identification information of the video stream to which each frame data belongs; intercepting any frame data of the target video stream as the input of the examination table detection model to obtain all the examination table target frames in the examination room; sorting all the examination table target frames according to the Y-axis order of the lower left corner coordinate to obtain the examination table target frame with the farthest distance on the Y-axis; calculating the position of the area where the examination table target frame with the farthest distance on the Y-axis is located in the monitoring image to determine whether the current camera is the left camera or the right camera. By acquiring the spatial position of the examination table as the relative reference system of the examination table target frame, combining the target tracking algorithm with the examinee-examination table matching algorithm to complete the separation of the examinee target frame in the pedestrian target frame, and then reversely separating the invigilator's target frame, the problem of being unable to distinguish between examinees and invigilators in the examination room monitoring scene is effectively solved, and more adequate preparation is provided for the analysis of abnormal behavior of intelligent inspection.
[0102] and Figure 1 For methods corresponding to Figure 6 , an embodiment of the present invention provides an electronic device, including:
[0103] at least one processor;
[0104] at least one memory for storing at least one program;
[0105] When the at least one program is executed by the at least one processor, the at least one processor implements the above method.
[0106] It can be seen that the contents of the above method embodiments are all applicable to the present system embodiments, the functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0107] In addition, an embodiment of the present invention further discloses a computer program product or a computer program, which is stored in a computer-readable storage medium. A processor of a computer device can read the computer program from a computer-readable storage medium, and the processor executes the computer program, so that the computer device executes the above method. Similarly, the contents of the above method embodiment are all applicable to the storage medium embodiment, and the functions specifically implemented by the storage medium embodiment are the same as those of the above method embodiment, and the beneficial effects achieved are also the same as those achieved by the above method embodiment.
[0108] It will be appreciated by those skilled in the art that all or some of the methods disclosed above and the system may be implemented as software, firmware, hardware and appropriate combinations thereof. Some physical components or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transient medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.
[0109] The above is a specific description of the preferred implementation of the present disclosure, but the present disclosure is not limited to the above-mentioned implementation mode. Technical personnel familiar with the field can also make various equivalent deformations or substitutions without violating the spirit of the present disclosure. These equivalent deformations or substitutions are all included in the scope defined by the claims of the present disclosure.
Claims
1. A method for visual detection of examinees and invigilators in an examination room, characterized in that: The method comprises the following steps: Acquire a target video stream, wherein the frame data of the target video stream carries a storage timestamp of each frame data and channel identification information of the video stream to which each frame data belongs; Intercept any frame of the target video stream as the input of the examination table detection model to obtain the target frames of all examination tables in the examination room; Sort all the test table target frames according to the Y-axis order of the lower left corner coordinates to obtain the test table target frame with the longest distance on the Y-axis; Calculate the Y-axis farthest distance test table target frame in the monitoring image to determine whether the current camera is a left camera or a right camera; The step of sorting all the test table target frames according to the Y-axis order of the lower left corner coordinates includes: Pairing the elements in the sorted and numbered set R in pairs along the Y axis in vertical order; The candidate test area set E is formed by matching the diagonal lines of the target box; After completing the examination table target detection task, execute the examination table target detection resource release operation; The intercepting any frame data of the target video stream comprises: Input the pedestrian detection model and target tracking algorithm to the target video stream frame by frame to obtain a pedestrian target frame set P with a tracking ID; traverse the set P to calculate the center point P1 of each pedestrian target frame; The method further comprises: Determine whether the center point P1 is located within an element in the candidate's test area set E; If the center point P1 is located in an element of the set E, then P1 corresponds to the candidate-associated frame number Sf+1 in the pedestrian attribute, and if Sf> the preset candidate threshold, the pedestrian is judged to be a candidate; If the center point P1 is not in the set E, then P1 corresponds to the number of proctor-associated frames Tf+1 in the pedestrian attribute. If Tf> the preset proctor threshold, the pedestrian is identified as a proctor. The step of judging whether the center point P1 is located within an element in the candidate's test area set E comprises: If the candidate has been identified, the bound candidate test area is retrieved to determine whether the candidate detection frame is still within the area. If so, the verification is passed. Otherwise, the candidate's pedestrian attribute information is destroyed and the relationship is re-established. If the invigilator has been identified, traverse the candidate test area set E to determine whether the invigilator detection box is still in the area. If so, Tf-1. If Tf<the preset invigilator threshold, destroy the invigilator's pedestrian attribute information and re-establish the relationship; if the pedestrian detection box is not in any element area in the set E, the verification is passed.
2. The method for visually detecting examinees and invigilators in an examination room according to claim 1, characterized in that: Intercepting any frame of the target video stream as the input of the examination table detection model to obtain all the examination table target frames in the examination room, and also including: Label all the target frames of the examination tables in the examination room to form a target detection frame set D; The set D of object detection frames of the examination tables is sorted and numbered, and its ID is fixed and matched with the seat number of the examination room.
3. The method for visually detecting examinees and invigilators in an examination room according to claim 2, characterized in that: Sorting and numbering the test table object detection frame set D includes: S1, traverse the test table target detection frame set D; S2, searching for the target frame box_rt with the smallest distance from the X=0 axis in the target detection frame set D of the examination table, and determining that the target frame is the first row and first column of the desk based on the prior rule; S3, searching for the target frame box_ld with the largest distance from the Y=0 axis in the target detection frame set D of the examination table, and determining that the target frame is the desk in the first row and the Nth column based on the prior rule; S4, constructing a straight line L based on the center points of the two boxes box_rt and box_ld, traversing the test table target detection box set D to filter out the target boxes that are penetrated by the straight line L to form a set R; wherein the set R currently stores the test table target boxes of the first row in the test room; S5, sorting the set R according to the size of the X-axis, and giving each frame a number in sequence; S6, calculating the size of the set R to obtain the number of candidates in the row; S7, removing all elements in the set R from the test table object detection frame set D; S8, if the test table object detection frame set D is an empty set, sort and number all the test tables to form the set R; If the examination table object detection frame set D is not an empty set, repeat the above steps until the examination table object detection frame set D is an empty set.
4. A visual detection system for examinees and invigilators in an examination room, characterized in that: include: An acquisition module is used to acquire a target video stream, wherein the frame data of the target video stream carries a storage timestamp of each frame data and channel identification information of the video stream to which each frame data belongs; The model input module intercepts any frame data of the target video stream as the input of the examination table detection model to obtain the target frames of all examination tables in the examination room; A sorting module sorts all the test table target frames according to the Y-axis order of the lower left corner coordinates to obtain the test table target frame with the longest distance on the Y-axis; A judgment module calculates the position of the area where the target frame of the examination table with the longest distance on the Y axis is located in the monitoring image and determines whether the current camera is a left camera or a right camera; The step of sorting all the test table target frames according to the Y-axis order of the lower left corner coordinates includes: Pairing the elements in the sorted and numbered set R in pairs along the Y axis in vertical order; The candidate test area set E is formed by matching the diagonal lines of the target box; After completing the examination table target detection task, execute the examination table target detection resource release operation; The intercepting any frame data of the target video stream comprises: Input the pedestrian detection model and target tracking algorithm to the target video stream frame by frame to obtain a pedestrian target frame set P with a tracking ID; traverse the set P to calculate the center point P1 of each pedestrian target frame; Also includes: Determine whether the center point P1 is located within an element in the candidate's test area set E; If the center point P1 is located in an element of the set E, then P1 corresponds to the candidate-associated frame number Sf+1 in the pedestrian attribute, and if Sf> the preset candidate threshold, the pedestrian is judged to be a candidate; If the center point P1 is not in the set E, then P1 corresponds to the number of proctor-associated frames Tf+1 in the pedestrian attribute. If Tf> the preset proctor threshold, the pedestrian is identified as a proctor. The step of judging whether the center point P1 is located within an element in the candidate's test area set E comprises: If the candidate has been identified, the bound candidate test area is retrieved to determine whether the candidate detection frame is still within the area. If so, the verification is passed. Otherwise, the candidate's pedestrian attribute information is destroyed and the relationship is re-established. If the invigilator has been identified, traverse the candidate test area set E to determine whether the invigilator detection box is still in the area. If so, Tf-1. If Tf<the preset invigilator threshold, destroy the invigilator's pedestrian attribute information and re-establish the relationship; if the pedestrian detection box is not in any element area in the set E, the verification is passed.
5. An electronic device, characterized in that: The electronic device comprises: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the method for visual detection of examinees and invigilators for examination room scenarios as described in any one of claims 1 to 3.
6. A computer-readable storage medium storing a program executable by a processor, characterized in that: The program executable by the processor is used to execute the method for visual detection of examinees and invigilators for examination room scenarios as described in any one of claims 1 to 3 when executed by the processor.
Citation Information
Patent Citations
Anti-cheating method, device and system for exam
CN108039029A
Video processing method and device and computer readable storage medium
CN116996740A
Examination room cheating behavior identification method and device based on seat identification algorithm
CN117351421A
Behavioural monitoring system
GB202210869D0
Object Behavior Analysis Method, Information Display Method, and Electronic Device
US20220319234A1
