In recent decades, people pay more and more attention to garbage disposal. Aiming at the problems of heavy workload and low efficiency of manual garbage classification, an intelligent garbage classification system based on machine learning is proposed. This design uses STM32F103ZET6 as the main control module and OpenMv4 H7 as the camera acquisition module to collect garbage pictures. BRISK feature extraction algorithm is adopted to realize image detection, and JQ8400 voice module is used to realize voice broadcast function. After testing, the correctness of the system's garbage classification reaches 98%, which can be categorized as kitchen waste, harmful waste, other waste and recyclable waste according to the current garbage classification scheme. The system is simple in operation, reliable in performance and low in cost.