A Posture-Based Exam Cheating Detection Method

Abstract

The invention discloses a test cheating detection method based on posture, comprising the following steps: image preprocessing: unifying the size of the image, performing illumination compensation on the image, and increasing the contrast of the image; extracting the posture feature of the person in the examination room video: through a deep neural network Obtain the coordinates of the key points of the person in the image and complete the splicing of the human body pose; the candidate pose sequence generation: use the method based on voting to determine the position of the candidate, and then use the Euclidean distance between the centroid of the candidate detected in the image and the position of the candidate as the judgment criterion , to generate the candidate's posture sequence in time; to identify the suspicious behavior of the candidate; to determine the cheating candidate and to intercept the cheating segment. The posture-based test cheating detection method can accurately and quickly screen out the suspected cheating fragments in the massive test monitoring video, so as to save manpower expenses, improve work efficiency and reduce the cost of video playback.

Description

technical field [0001] The invention relates to a method for detecting cheating in exams. Firstly, the posture features of people in the video are extracted, and then the posture features are statistically analyzed to determine whether candidates have cheating behavior. It is suitable for various examination video analysis and belongs to computer vision and video understanding technology. field. Background technique [0002] In order to analyze the test style problems in the test, such as the cheating behavior of the candidates and the inaction of the invigilator, etc., after various tests such as the college entrance examination, the postgraduate examination, the adult self-examination and the academic level test, it is necessary to replay the examination video. Exam video playback tasks require a lot of time and money costs, especially for large exams. Large-scale exams have dozens of test sites and hundreds of exam rooms, and the exam takes a long time (usually more than...

AI Technical Summary

Problems solved by technology

This method cannot conduct a comprehensive analysis of the cheating behavior of candidates

[0006] Publication number CN201910110600 patent "an anti-cheating monitoring system in examination room" uses smart wearable devices, camera modules, positioning modules and monitoring modules to simultaneously detect the heart rate and behavior characteristics of candidates, but such a huge system is not only expensive but also extremely cumbersome to use. almost impossible to promote

[0007] Patent Publication No. CN201010226332 "Intelligent Video Recognition Method for Examination Cheating Events" first extracts the foreground and background of the examination room through the establishment and update of the Gaussian mixture model for the monitoring video of the examination room, and extracts the motion pixel feature value according to the foreground image, and then The eigenvalues ​​of the area where candidates are located in the examination room are extracted and used as the data for the next decision tree judgment, and the corresponding cheating events are judged based on the data of the decision tree, but the system is somewhat one-sided in the detection of cheating behavior of candidates

Images