Method for detecting abnormity of automobile printed circuit board in industrial production process

The invention provides an automobile printed circuit board anomaly detection method in the industrial production process, and belongs to the technical field of image processing. According to the method, an encoder-decoder structure is adopted, and a model is trained on a normal printed circuit board image; the method specifically comprises the following steps that model input is a single printed circuit board image, and firstly, an encoder is adopted to extract image features through a heavy parameter structure; then, a bottleneck module compresses the feature map by adopting a cross attention series structure; the decoder adopts a convolution module to regenerate a printed circuit board image according to the feature map; and finally, the detection head obtains an abnormal probability graph by comparing the difference between the input image and the generated image. According to the invention, the model is trained only by normal samples, effective detection of the abnormal printed circuit board is realized,