Thanks for your great work! I have some problems when I try to ues your dataset, and hope to get your help!
I find that in the /annotations/DEX_YCB_s0_train_data.json, there is also mano_param for every image. I wonder whether they are in camera frame or world frame. If I want to use the hand parameters in this json file, how can I get the parameters of the hand translation(1X3) and the object rotation and translation(4X4)?
I tried to visualize the hand pose using the paramaters both in the label file and the json file, but I found neither of them was right and they were different from each other. I couldn't figure out the reason. In my opinion, even if the paramaters of the object translatation I used were wrong, the hand pose at least should be right. So the result really confused me.
This is out of context of this repo. DEX_YCB_s0_train_data.json seems to be created by the authors of the HandOccNet repo. You may want to seek answers there.
Thanks for your great work! I have some problems when I try to ues your dataset, and hope to get your help!
/annotations/DEX_YCB_s0_train_data.json
, there is alsomano_param
for every image. I wonder whether they are in camera frame or world frame. If I want to use the hand parameters in this json file, how can I get the parameters of the hand translation(1X3) and the object rotation and translation(4X4)?Thanks again and looking forward to your reply!