Data processing of SURREAL and H3D

[hnuzhy]

Hello, I want to ask some data processing details of the RegDA algorithm. https://github.com/thuml/Transfer-Learning-Library/tree/master/examples/domain_adaptation/keypoint_detection

As shown in the README, datasets SURREAL and Hand-3d-Studio can be downloaded automaticly with processd, sampled and cropped versions which are provided by you.

• processd SURREAL dataset: https://github.com/thuml/Transfer-Learning-Library/blob/d950c3115557423cd5dcec3fdeee2c3ffb2cae5a/tllib/vision/datasets/keypoint_detection/surreal.py
• processd Hand-3d-Studio dataset: https://github.com/thuml/Transfer-Learning-Library/blob/d950c3115557423cd5dcec3fdeee2c3ffb2cae5a/tllib/vision/datasets/keypoint_detection/hand_3d_studio.py

Now, I want to get the corresponding original image and labels of each processd sample. So, could you please release the cropping scripts for SURREAL and Hand-3d-Studio datasets?

[thucbx99]

The original data processing code is a bit mass. After some restructure, we will upload these scripts.