基于深度学习与视觉算法的垃圾分类装置设计OACSTPCD

In order to cope with the increasingly severe challenge of waste classification,this paper designs a waste sorting device based on machine learning and vision algorithms.The device collects data from the video camera and sorts them by the Maix bit core board,in which the machine learning network MobileNet and the normalization algorithm in the vision module are applied to increase the sorting accuracy dramatically.After classification,the Maix bit core board communicates with the STM32F103 board to control the corresponding servo and the speech announce system to implement the sorting operation for the relevant type of waste.After experimental validation,this device can achieve garbage sorting speeds of less than 1 second and accuracy greater than 98％.Therefore,it possesses strong prac-ticality in real applications.