NeuroCam: A Neuromorphic Intelligent Camera System for Face Recognition

Neuromorphic chips based on spiking neural networks achieve highly integrated functions of information perception, processing, and learning by mimicking the compact human brain cortex leveraging spatiotemporally sparse spikes towards efficient sensory data representation, transmission and processing. They significantly reduce the computational loads of intelligent systems with improved energy efficiency. In this work, we have developed a low-power and low-cost intelligent camera for human face recognition based on our previously proposed neuromorphic chip. The system could accurately and promptly accomplish the face recognition task at a frame rate of 12.5 fps while consuming a moderate amount of 5.88W power. It realized a 100% precision rate and a 97.50% recall rate.