Train VAE for image generation

I am trying to implement the code given in the following link with my own dataset containing RGB images of size 200*200:

https://in.mathworks.com/help/deeplearning/ug/train-a-variational-autoencoder-vae-to-generate-images.html

I am getting the following errors in the Train model part:

I can make out that there is some issue happening with the custom functions ('modelLoss', 'elboLoss') given in this link. But I am unable to understand how can I resolve these issues. Someone please kindly guide me . I will be thankful to you.

Answer 1

[Ranjeet]

Hi Debojit,

As per the error message provided, the decoder (netD) output size is not matching with what is being fed to the encoder (netE) input.

I suggest analysing the networks (can refer to this resource - analyzeNetwork) and check if the decoder output is same to that of encoder input (image of size 200*200).

Also, size of ‘Y’ and ‘T’ can be cross-checked to be equal in ‘elboLoss’ function in the referred example – Train Variational Autoencoder (VAE) to Generate Images