Mask-Robust Face Verification for Online Learning via YOLOv5 and Residual Networks

TL;DR

The paper tackles identity verification in online learning environments by integrating YOLOv5-based face detection with a ResNet-based feature extractor to verify student identities via Euclidean distance against a stored embedding database. It introduces a mask-robust recognition pipeline trained on a proprietary dataset, achieving a detector precision of 0.83494 and recall of 0.81308, and a ResNet-34 based 128D embedding extractor with competitive performance on the LFW benchmark (mean error 0.993833, SD 0.00272732). An end-to-end system with a PyQt GUI and a student face feature database supports real-time authentication, enabling secure access to cloud classrooms. The approach demonstrates practical impact for online education by improving attendance tracking and reducing impersonation risk, with potential for deployment in online learning platforms and future enhancements through dataset expansion and optimization.

Abstract

In the contemporary landscape, the fusion of information technology and the rapid advancement of artificial intelligence have ushered school education into a transformative phase characterized by digitization and heightened intelligence. Concurrently, the global paradigm shift caused by the Covid-19 pandemic has catalyzed the evolution of e-learning, accentuating its significance. Amidst these developments, one pivotal facet of the online education paradigm that warrants attention is the authentication of identities within the digital learning sphere. Within this context, our study delves into a solution for online learning authentication, utilizing an enhanced convolutional neural network architecture, specifically the residual network model. By harnessing the power of deep learning, this technological approach aims to galvanize the ongoing progress of online education, while concurrently bolstering its security and stability. Such fortification is imperative in enabling online education to seamlessly align with the swift evolution of the educational landscape. This paper's focal proposition involves the deployment of the YOLOv5 network, meticulously trained on our proprietary dataset. This network is tasked with identifying individuals' faces culled from images captured by students' open online cameras. The resultant facial information is then channeled into the residual network to extract intricate features at a deeper level. Subsequently, a comparative analysis of Euclidean distances against students' face databases is performed, effectively ascertaining the identity of each student.

Figures (10)

• Figure 1: The framework of student authentication system. It consists of a head section for preprocessing, a backbone section and a SPPF section for locating face coordinates, and a ResNet-29 section for extracting high-dimensional face features.
• Figure 2: Feature maps of varying sizes.
• Figure 3: Feature extraction structure.
• Figure 4: Precision for YOLOv5s.
• Figure 5: Recall for YOLOv5s.