Machine learning based inference privacy sanitization for online proctoring
The detection of cheating behavior is the key direction of online proctoring research, but the privacy protection of candidates is often neglected, resulting in the risk of personal information leakage of candidates in proctored video streams. In recent years, how to sanitize the privacy in online i...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis-Master by Coursework |
| Language: | English |
| Published: |
Nanyang Technological University
2022
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/157416 |
| _version_ | 1811686381371523072 |
|---|---|
| author | Chen, Xinyu |
| author2 | Tay Wee Peng |
| author_facet | Tay Wee Peng Chen, Xinyu |
| author_sort | Chen, Xinyu |
| collection | NTU |
| description | The detection of cheating behavior is the key direction of online proctoring research, but the privacy protection of candidates is often neglected, resulting in the risk of personal information leakage of candidates in proctored video streams. In recent years, how to sanitize the privacy in online invigilation videos without affecting the accuracy of examinee identity and cheating behavior identification has become a very important research topic.
In this work, we have studied the current advanced privacy protection methods, combined with the particularity of online proctoring scenarios, and proposed new online exam cheating detection process, using high-resolution background matting and human body key point recognition method to complete the protection of candidates' privacy. In addition, we apply a state-of-the-art AlphaAction model to complete the detection of cheating behavior. Then we test the above methods comprehensively by using our own recorded videos of simulated online exam cheating. The final experimental evaluation results show that the privacy protection method adopted in this dissertation is effective and will not affect the accuracy of subsequent cheating detection, and it also provides valuable insight for future study on online proctoring privacy protection. |
| first_indexed | 2024-10-01T04:59:31Z |
| format | Thesis-Master by Coursework |
| id | ntu-10356/157416 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T04:59:31Z |
| publishDate | 2022 |
| publisher | Nanyang Technological University |
| record_format | dspace |
| spelling | ntu-10356/1574162023-07-04T17:50:36Z Machine learning based inference privacy sanitization for online proctoring Chen, Xinyu Tay Wee Peng School of Electrical and Electronic Engineering wptay@ntu.edu.sg Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision The detection of cheating behavior is the key direction of online proctoring research, but the privacy protection of candidates is often neglected, resulting in the risk of personal information leakage of candidates in proctored video streams. In recent years, how to sanitize the privacy in online invigilation videos without affecting the accuracy of examinee identity and cheating behavior identification has become a very important research topic. In this work, we have studied the current advanced privacy protection methods, combined with the particularity of online proctoring scenarios, and proposed new online exam cheating detection process, using high-resolution background matting and human body key point recognition method to complete the protection of candidates' privacy. In addition, we apply a state-of-the-art AlphaAction model to complete the detection of cheating behavior. Then we test the above methods comprehensively by using our own recorded videos of simulated online exam cheating. The final experimental evaluation results show that the privacy protection method adopted in this dissertation is effective and will not affect the accuracy of subsequent cheating detection, and it also provides valuable insight for future study on online proctoring privacy protection. Master of Science (Signal Processing) 2022-05-12T02:56:17Z 2022-05-12T02:56:17Z 2022 Thesis-Master by Coursework Chen, X. (2022). Machine learning based inference privacy sanitization for online proctoring. Master's thesis, Nanyang Technological University, Singapore. https://hdl.handle.net/10356/157416 https://hdl.handle.net/10356/157416 en application/pdf Nanyang Technological University |
| spellingShingle | Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision Chen, Xinyu Machine learning based inference privacy sanitization for online proctoring |
| title | Machine learning based inference privacy sanitization for online proctoring |
| title_full | Machine learning based inference privacy sanitization for online proctoring |
| title_fullStr | Machine learning based inference privacy sanitization for online proctoring |
| title_full_unstemmed | Machine learning based inference privacy sanitization for online proctoring |
| title_short | Machine learning based inference privacy sanitization for online proctoring |
| title_sort | machine learning based inference privacy sanitization for online proctoring |
| topic | Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision |
| url | https://hdl.handle.net/10356/157416 |
| work_keys_str_mv | AT chenxinyu machinelearningbasedinferenceprivacysanitizationforonlineproctoring |