I think you can download ITK-SNAP to check the visualization results. In my experience, maybe IRM training fails. Instead of predicting residuals, IRM output noisy or zero images. The main reason is that the two-stage network of this paper (CPM&IRM) sharing the same gradient has the problem of training instability.
The solution is to check visualisation results as soon as possible. In my experience, training problems are often spotted early in the training phase with checkpoints 40,000 iters. If there is a model collapse, it is sufficient to train from scratch in time.
Originally posted by @wyhlovecpp in #11
Hello, I get the same result with the issue #11 , I have tried about 10 more training tasks, but all of them get bad result. they have PSNR about 20 at step 40000, but only about 22 at step 500000, is the probability of model collapse really that high? the image behind is our result, which is {input, gt, coarse, residual, final}
Originally posted by @wyhlovecpp in #11
Hello, I get the same result with the issue #11 , I have tried about 10 more training tasks, but all of them get bad result. they have PSNR about 20 at step 40000, but only about 22 at step 500000, is the probability of model collapse really that high? the image behind is our result, which is {input, gt, coarse, residual, final}