In order to monitor the surface condition of the conveyor belt in the process of running, a method of beam structure light irradiation based on machine vision is adopted. A spring-type mechanical vibration damping device is designed to improve the focusing quality of the camera, and an algorithm is proposed to solve the selection of spring parameters under different flutter amplitudes. Yolov7 deep learning algorithm was adopted and ACmix attention mechanism was introduced to identify the surface cracks of conveyor belt. The experimental results show that the improved YOLOV7-ACmix algorithm can effectively improve the accuracy and generalization ability of image recognition.