Automobile gear defect identification method based on machine vision

The invention provides an automobile gear defect identification method based on machine vision, and relates to the field of computer vision. The invention provides an automobile gear defect identification method. The method comprises the steps of constructing a machine vision detection module, preparing an automobile gear defect data set, training the automobile gear defect data set by using a machine vision algorithm and detecting automobile gear defects by using a detection model. Meanwhile, a machine vision detection module is put forward in a machine vision algorithm, a self-attention mechanism, a space attention mechanism and a channel attention mechanism are fused, the attention mechanisms are subjected to operations of convolution, element-by-element addition and the like, feature map weights are properly fused, and the feature map weight is calculated. Therefore, the recognition effect of machine vision in an automobile gear defect recognition scene is improved. 本发明提出了一种基于机器视觉的汽车齿轮缺陷识别方法，涉及计算机视觉领域；本发明