Emotion recognition system based on artificial intelligence optoelectronic technology
Abstract
In the context of psychological care for adolescents and safety monitoring for the elderly, traditional emotion monitoring devices face challenges such as low accuracy in single-modal recognition, cloud data leakage, and high hardware costs. This paper designs an emotion recognition system based on artificial intelligence optoelectronic technology. The system employs an ESP32-S3 main control unit paired with a K230 computing chip in a heterogeneous collaborative architecture, integrating visual optical flow imaging, voice, and physiological signal multimodal optoelectronic perception solutions. By completing lightweight YOLO model local deployment through INT8 quantization, all sensitive data is processed offline at the edge and stored securely, accompanied by a three-tier progressive emotional intervention mechanism. This solution balances low cost, privacy security, and real-time response capability, providing a feasible approach for optoelectronic edge intelligent devices in the livelihood sector.
