Transfer Learning in Detecting E-Assessment Malpractice from a Proctored Video Recordings.

Abstract

E-assessment Malpractice Image classification is very important in education applications to develop an effeactive studies. In this paper, we use imagenet benchmark dataset to classify five types of examination taking and malpractices involved (a) cheat from text books/notes/papers. (b) Using a phone to call. (c) asking a friend in the test room (d) operating a phone and (e) a normal exam taking without cheating. Due to the small number of training dataset, our classification systems evaluate deep transfer learning for feature extraction. . During these exams it is difficult to keep track of every student’s screen at the same time to check if anyone is showing fraudulent behaviour. Even when recording all students’ activities during exam and watching it afterwards to depend if they cheated is very labour intensive. This thesis uses a special type of Convolutional neural network called Inception which is are widely known for their effectiveness in image classification on students’ proctored video recordings to determine if they show any malpractice behaviors, allowing us to build a framework to automate this labour intensive process system. The objective of this study is to increase the classification accuracy, speed the training time and avoid the overfitting. In this study, we trained our architecture to involve minimal pre-processing for 30-epoch number in order to study its impact on classification performance and consuming time. In addition, the paper benefits acceptable results with small number of epoch in limited time. Our interpretations confirm that transfer learning provides reliable results in the case of small dataset. The proposed system outperforms the state-of-the-art methods and achieve 96.8% classification accuracy.