3D Articulated Hand Pose Estimation with Single Depth Images

D. Tang, H.J. Chang*, A. Tejani*, T-K. Kim
Latent Regression Forest: Structured Estimation of 3D Hand Posture, Proc. of IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), Columbus, Ohio, USA, 2014 (oral, accept rate=5.75%).
*indicates equal contribution.

Download: PDF [ 960 KB] [Demo on YouTube video] [Project page]
D. Tang, T.H. Yu and T-K. Kim
Real-time Articulated Hand Pose Estimation using Semi-supervised Transductive Regression Forests, Proc. of IEEE Int. Conf. on Computer Vision (ICCV), Sydney, Australia, 2013 (oral, accept rate=2.7%).

Download: PDF [ 1,465 KB] [Demo on YouTube video] [Slides]
Download: Training / Testing / Our results

Label description:
  • Each line is corresponding to one image.
  • Each line has 16x3 numbers, which indicates (x, y, z) of 16 joint locations. Note that these are joint CENTRE locations.
  • Note that (x, y) are in pixels and z is in mm.
  • The order of 16 joints is Palm, Thumb root, Thumb mid, Thumb tip, Index root, Index mid, Index tip, Middle root, Middle mid, Middle tip, Ring root, Ring mid, Ring tip, Pinky root, Pinky mid, Pinky tip.
  • We used Intel Creative depth sensor. Calibration parameters can be obtained as in Page 119 of SDK Manual:

Many thanks to Guillermo Garcia for help with publishing this dataset.
This project was supported by the Samsung Advanced Insititute of Technology(SAIT).