Adrian Bulat

@ducha-aiki, Did you had any look with the training of ResNet on caffe? Thanks

@ducha-aiki thanks a lot for your prompt answer and for the recommended paper. I saw in your implementation that you are not using the Scale layer to learn the parameters,...

@ducha-aiki makes sense :) Thanks again for your help!

Hi @ablavatski, I don't remember the exact range but at that point expect values in the range [-70, 70] approximatively. The depth predicted should be consistent with the one from...

Hi @wqz960, We only evaluate on AFLW2000-3D, 300W, 300VW and LS3D-W-Balanced. AFLW2000-3D simple represent the first 2000 images of the AFLW dataset that were re-annotated Zhu et al. Thanks, Adrian

Hi @ouceduxzk thanks a lot for your interest and kind words! Unfortunatelly at this moment there is no possiblity to run it on a CPU, because there is no cpu...

Thanks a lot! I will have a look tomorrow on it.

Hi @leiwang001 the short answer is no. The longer is yes :) The code performs single person pose estimation, however if you use a different method to get the locations...

Unfortunatelly there is no imediat plan for this at the moment. However, converting between the two should be straightforward so I am encouraging you to give it a try.

@pixelou thanks a lot for your contribution, it will definitely come in handy. If you want to make a pull request I am happy to add it to the instructions.