search

hqng / imm-pytorch

PyTorch implementation of "Unsupervised Learning of Object Landmarks through Conditional Image Generation", Tomas Jakab*, Ankush Gupta*, Hakan Bilen, Andrea Vedaldi, Advances in Neural Information Processing Systems (NeurIPS) 2018.
15 stars 2 forks source link

Reconstruction results problem #1

Open prashastbindal opened 4 years ago

prashastbindal commented 4 years ago

Hi hqng,

First of all thank you for doing this implementation, it's a great help to us! I recently tried recreating the results on Celeb A dataset, but I am unable to achieve results as good as the original research paper. Were you able to recreate the original paper results with your training? If so, would it be possible for you to share a pretrained model that you might have?

I am sharing a few screenshots from my run fig_108




I also ran the model on Human3.6M and the following are my results fig_8 :

josesosajs commented 3 years ago

Hi @prashastbindal, I'm facing the same problem, do you have any updates on this?