Real-Time Face Detection and Recognition using Deep Learning and Nvidia Optimised Models

This study explores the application of advanced machine learning (ML) techniques to real-time face detection and recognition using the NVIDIA Jetson Nano platform. Utilizing a pre-trained ResNet-18 model adapted for facial recognition tasks with transfer learning, the system integrates NVIDIA’s optimized Docker containers to enhance computational efficiency and reliability in a variety of operational environments. The core methodology is comprised of real-time processing of video streams to detect and recognize faces, leveraging deep learning optimizations such as TensorRT and CUDA for low latency performance. This system aims to provide a scalable solution for automated attendance and surveillance applications, with a focus on accurate performance in dynamic settings. By addressing both technological and practical aspects, the research contributes to ongoing work in enhancing edge-based AI applications and offers a case study of their deployment in real-world scenarios.