Method for detecting abnormity of feeding tray in industrial production process

The invention provides an industrial production process feeding tray anomaly detection method, which comprises the following steps that: a model takes a single-frame image as input, extracts features through an encoder, fuses multi-scale feature maps through a bottleneck module, calculates the similarity of the feature maps and memory vectors one by one through a memory module, and weights the feature maps by taking the maximum similarity, so that the anomaly detection accuracy of the feeding tray is improved. The decoder performs multi-scale reconstruction on the weighted feature map by using pixel recombination, the detection head calculates the similarity of the corresponding positions of the output feature maps of the encoder and the decoder, and the similarity maps are fused to obtain the abnormal score map of the input image, so that the problem of the traditional reconstruction method in distinguishing normal and abnormal data is solved, and the accuracy of the abnormal data is improved. An unsupervise