Open yelusaleng opened 2 years ago
Hello again,
After re-reading the article, you're right, they have been trained separately on different datasets before being merged.
@RonyAbecidan lol, thanks again. I would like to know the model details, do you know what is the full model for both networks? I was able to successfully train the mantranet using your code, but the results are not so good, and I would like to train the two models correctly and separately.
after re-reading the paper, I guess that the image manipulation tract feature extractor is the same as VGG net. but, I still do not know the structure of the second network.
hi, Are the two networks (image manipulation tract feature extractor and local anomaly detection network) trained separately?