CyberProctor engineers have leveraged the accuracy of Facial Recognition in their Online Proctoring algorithm. Mistaken identity and students substituting for others is virtually impossible, even among identical twins. The technologies used include:
Deep Learning
Deep Learning algorithms are revolutionizing the Computer Vision field, capable of obtaining unprecedented accuracy in Computer Vision tasks, including Image Classification, Object Detection, Segmentation, and more.
Face Applications
Computer Vision algorithms can be used to perform face recognition, enhance security, aid law enforcement, detect tired, drowsy drivers behind the wheel, or build a virtual makeover system.
So, how does deep learning + face recognition work?
The secret is a technique called deep metric learning and uses a complex vector calculation to train the network on 128 points on each users face. These common points are converted to histograms, serialized and saved into our data-set for comparison to streams and images. The identified image/stream is labeled with the users unique key and recorded. We then use our proprietary motion detection sub-routine to record suspicious movements to alert staff and users.
When a student signs up for CyberProctor, and every time thereafter, a current picture as well as a current photo of their id are required and are saved in the users embedding file to improve accuracy. Once the network has been trained, proctoring can begin.