Closed mkocabas closed 5 years ago
Since you have original dataset, you can generate your own cache file by setting different sampling strategy. See _sample_dataset in hm36.py, it provides several sampling method.
Thanks for the answer @lck1201, by the time I already implemented it. I can get similar reconstruction error now.
@mkocabas Hi I am trying to reproduce the paper on standard h36m dataset Could you tell me how do you prepare dataset?
Hi @JimmySuen,
Great work, thanks for sharing! I wonder how can we modify the images of standart Human3.6M dataset (not the challenge one) to train with your model? I've the videos and images of the original dataset, but I couldn't find a script to modify image paths according to your dataset caches.