How to change image size to, say 128?

In "openExample('nnet/TrainGenerativeAdversarialNetworkGANExample')", the image size is set to 64. When changed all "found" places to 128, it showed a non-scaler error.

Please help to resolve this problem.

Thank you!

Answer 1

[Image Analyst]

I think most (or all?) network architectures are built to expect a certain size image upon input. I don't think you can just change the size to whatever you want.

[John]

Added an according layer to both generator and discriminator, it worked up to 5 iterations (still in epoch 1), then it shows an error "state must be a table, value must be 1-by-1 cell array or dlarray.

Displayed "dlnetGenerator.State", it does show the correct table with 1-by-1 dlarray.

Comparing to 64 image, it has one extra layer with higher number.

It would be an excellent tutorial for all. Hope experts can help.

Thanks.