E-portal: A Smart Computer Vision System for Face Detection and Recognition in Residential Buildings

Existing smart surveillance systems are widespread due to the popularity of artificial intelligence solutions. These systems are applicable in different environments, particularly in Residential Buildings (RBs). However, RBs pose several challenges, including infrastructural constraints, such as grayscale footage, low frame rates, outdated, and low-capacity hardware. These challenges remain unaddressed. To address these shortcomings, this paper proposes a smart, IoT-integrated surveillance system in RBs that combines sensor-based hardware aligned with dual-stage face recognition for low-resource environments. The proposed system integrates YOLOv8n for face detection chosen for its ability to deliver accurate results without overwhelming limited hardware and VGG19 for face recognition. It achieves an overall accuracy of 96.88% while significantly reducing computational overhead. It is optimized for edge computing and low-specification surveillance setups, offering a scalable and reliable solution that mitigates hardware constraints.