The dataset was captured by a Kinect device. The images were collected under 2 different scenes. The first scene contains 1-2 persons with a variety of poses, the second scene contains 2-4 persons. There are occlusions in the dataset.
Example images:
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
The dataset contains two parts:
(a) original depth matrix data in .txt format, the resolution is 480x640.
scene_1 (6.2M) scene_2 (6.5M)
(b) the ground truth of the person's head location in image coordinates (x,y).
For each frame, the first row is the name of the correspond depth image, the number in the second row indicates the number of persons in the image, each following row contains the location of the center of each person's head, as (x,y).
scene_1_truth (4.8KB)
If you are considering using of the UTKinect-HumanDetection dataset, please cite the following reference.
@inproceedings{xia2011human,
title={Human detection using depth information by Kinect},
author={Xia, L. and Chen, C.C. and Aggarwal, JK},
booktitle={Computer Vision and Pattern Recognition Workshops (CVPRW), 2011 IEEE Computer Society Conference on},
pages={15--22},
year={2011},
organization={IEEE}
}
If you find any errors or problems, please report to Lu Xia