obman_train
obman_train copied to clipboard
hassony2
[cvpr19] Demo, training and evaluation code for generating dense hand+object reconstructions from single rgb images
Hi, I want to only use hand mano params to train my network. I visualized the GT and found that there is a misalignment. Is this because of the global_orient...
Thanks for open sourcing this work. I am trying to run a demo on a single image using the command: ```python image_demo.py --resume release_models/obman/checkpoint.pth.tar``` However, I get the following error:...
I train this model in HO3D dataset,but I get terriblr result,can some one tell me how to train,can get result as the figure 
For the webcam demo, the type for the 2d joint line coordinates aren't casted to integers, leading to an error. Pull request #30 created.
Hi @hassony2, So sorry about the changes in README for #30! This one should only be the necessary changes. Another note, the joint visualization seems to be inverted along the...
Thanks for your great work. I have several questions about training using Obman. I have just run `python traineval.py --atlas_predict_trans --atlas_predict_scale --atlas_mesh --mano_use_shape --mano_use_pca --freeze_batchnorm --atlas_separate_encoder`. After training for 200...
There are some pkl files in the meta folder of the dataset which contain some information of the corresponding sample. I found that there is a parameter named 'hand_pose', but...
Hi, I'm sorry to bother you again. We are studying the hand pose estimation and want to compare the results on ObMan with your method. So we wish to get...
Hi, I ran the command `python traineval.py --atlas_predict_trans --atlas_predict_scale --atlas_mesh --mano_use_shape --mano_use_pca --freeze_batchnorm --atlas_separate_encoder` and got the following results. i'm wondering if it's trained correctly. It seems like epe_mean or...
