Deep Knowledge

@QQR1 please have a look https://github.com/xiahaifeng1995/PaDiM-Anomaly-Detection-Localization-master/issues/8#issuecomment-806567383

You can use it directly

The size could be any value that is divisible by 16. as @harewei mentioned, upsampled output image matches the size of your input image ` score_map = F.interpolate(dist_list.unsqueeze(1), size=(x.size(2),x.size(3)), mode='bilinear',...

No, it is not normal,

Yes, please have a look on the **Inference time issue** https://github.com/xiahaifeng1995/PaDiM-Anomaly-Detection-Localization-master/issues/8#issuecomment-1002495651

Read the whole issue

it seems we have to create a black mask for each good images (in ground_truth)