JTan

I've done this for a custom dataset. Here's a simplified version of what I used below. I don't remember all the details as well but essentially you discretize samples from...

Thanks for the confirmation, this is slightly puzzling as the authors of beta-TCVAE report MIG scores of O(0.1) in their paper. Other third-party implementations also report MIG of O(0.1), e.g....

After some digging I am getting better results using the author's MIG calculation code - around 0.3-0.8 for most of my trained models. Perhaps the problem lies in shuffling the...

Yeah, that's what I mean about the shuffling, thanks for confirming, so that is not the cause of it. I am still confused because AFAIK this repository and the author's...

Hello, 1. The DCGAN framework is an arbitrary choice, I have used it before in WGAN with reasonable results, but you may have success in trying other frameworks, e.g. in...

To produce the images in their paper, the authors train ADE20k for 50 epochs using the semantic label maps as additional information. I haven't tried training on ADE20k yet, because...

Yes, I think implementing the VGG perceptual loss may help. Unfortunately I am quite busy at the moment, but it is top of the to-do list.

Hi Yanfei, I ran on a 16 core machine with a single Nvidia P100 GPU. Cheers, Justin On Tue, 9 Jul 2019 at 8:39 pm, Zhang Yanfei wrote: > Hi...

You should have the input image as a JPEG or PNG instead of BMP or GIF.

Hey, It’s from section 4.1 of the paper - concatenation of the representation with noise drawn from a fixed prior to form some latent vector Z. Cheers, Justin On Tue,...