EgoDexter

A Benchmark Dataset for Hand Tracking in Presence of Occlusions and Clutter

F. Mueller, D. Mehta, O. Sotnychenko, S. Sridhar, D. Casas, C. Theobalt
Real-time Hand Tracking under Occlusion from an Egocentric RGB-D Sensor
International Conference on Computer Vision (ICCV) 2017, Venice, Italy.

EgoDexter is an RGB-D dataset for evaluating algorithms for hand tracking in the presence of occlusions and clutter. It consists of 4 sequences with 4 actors (2 female), and varying interactions with various objects and cluttered background. Fingertip positions were manually annotated for 1485 out of 3190 frames. This dataset accompanies the ICCV 2017 paper, Real-time Hand Tracking under Occlusion from an Egocentric RGB-D Sensor.

License

This dataset can only be used for scientific/non-commercial purposes. Please refer to the detailed license which is also enclosed in the download file. If you use this dataset, you are required to cite the following paper. BibTeX, 1 KB

@inproceedings{OccludedHands_ICCV2017,
 author = {Mueller, Franziska and Mehta, Dushyant and Sotnychenko, Oleksandr and Sridhar, Srinath and Casas, Dan and Theobalt, Christian},
 title = {Real-time Hand Tracking under Occlusion from an Egocentric RGB-D Sensor},
 booktitle = {Proceedings of International Conference on Computer Vision ({ICCV})},
 url = {http://handtracker.mpi-inf.mpg.de/projects/OccludedHands/},
 numpages = {10},
 month = October,
 year = {2017}
}

Downloads

  • Compressed Zip: Single file (zip, 1.99 GB), SHA-256:
    7b7f1b357e9e1ee39b8ac11dcd870128cae69ae2fd15d0ae4aa7000f1eed164c
  • Browse: here

Data

  • Color: Intel RealSense SR300 @640x480 px
  • Depth: Intel RealSense SR300 @640x480 px
  • Color on Depth: Constructed from the depth and color image using the Intel RealSense SDK
  • Ground Truth: Manually annotated on depth data for 3D fingertip positions
  • Camera Calibration: For mapping between world and camera coordinate systems (depth, color)

Evaluation Metric

Since commonly used in hand tracking research, we compute the mean Euclidean error in 3D per frame. Average errors per sequence as well as percentage of frames below an error threshold can be found in the accompanying paper and in this file (which is also included in the download file).

Sequence Details

Please click the links below for a video preview of each sequence.
  1. Rotunda: Male user 01 in a living room environment.
  2. Desk: Female user 01 in an office environment.
  3. Kitchen: Female user 02 in the kitchen.
  4. Fruits: Male user 02 interacting with fruits.

Contact

Franziska Mueller
frmueller@mpi-inf.mpg.de

Back to ICCV 2017