Lie-X: Depth Image Based Articulated Object Pose Estimation, Tracking, and Action Recognition on Lie Groups
Chi Xu , Lakshmi Narasimhan Govindarajan, Yu Zhang, Li Cheng (Project Leader)
Introduction
Pose estimation, tracking, and action recognition of articulated objects from depth images are important and challenging problems, which are normally considered separately. In this paper, a unified paradigm based on Lie group theory is proposed, which enables us to collectively address these related problems. Our approach is also applicable to a wide range of articulated objects. Empirically it is evaluated on lab animals including mouse and fish, as well as on human hand. On these applications, it is shown to deliver competitive results compared to the state-of-the-arts, and non-trivial baselines including convolutional neural networks and regression forest methods.
Video
Datasets
- Our fish pose estimation dataset (please contact Li or Lakshmi.)
- Our fish pose estimation result on the above dataset (unit 10um): lie_fish_jnts_estm_result.txt
- Our fish action dataset: please download FishActionData.zip
- Our mouse pose estimation dataset: please download mouse_dataset.zip
- Our mouse pose estimation result on the above dataset: lie_mouse_jnts_estm_result.txt
- NYU hand pose estimation dataset
- Our hand pose estimation result on NYU dataset: lie_hand_jnts_estm_result.txt
Executable
The executable of Lie-X for mouse can be downloaded here. Note that it is an alpha version which would be updated in the near future.
Publications
Chi Xu, Lakshmi Narasimhan Govindarajan, Yu Zhang, Li Cheng. Lie-X: Depth Image Based Articulated Object Pose Estimation, Tracking, and Action Recognition on Lie Groups. In International Journal of Computer Vision (IJCV), in press, 2016. [pdf]