###############################
### GANerated Hands Dataset ###
###############################

----------------------------------------------------------------------------------------------------------------------------------------------
|  Terms of use:                                                                                                                             |
|  The provided dataset is intended for research purposes only and any use of it for non-scientific and/or commercial means is not allowed.  |
|  This includes publishing any scientific results obtained with our data in non-scientific literature, such as tabloid press.               |
|  If you use this dataset, you are required to cite the following paper:                                                                    |
|                                                                                                                                            |
|  Mueller F, Bernard F, Sotnychenko O, Mehta D, Sridhar S, Casas D, Theobalt C.                                                             |
|  GANerated Hands for Real-Time 3D Hand Tracking from Monocular RGB.                                                                        |
|  Proc. of IEEE Computer Vision and Pattern Recognition 2018.                                                                               |
|                                                                                                                                            |
|  Refer to the license (license.txt) distributed with the data.                                                                             |
|                                                                                                                                            |
----------------------------------------------------------------------------------------------------------------------------------------------

frames without object: 143,449
frames with object: 188,050

camera intrinsics:
[ 617.173,       0, 315.453,
        0, 617.173, 242.259,
        0,       0,       1]

Directory structure:
* joints.png
* license.txt
* README.txt (this file)
* data/
	* withObject / noObject
		* <partition>/ (s.t. every directory contains 1024 example frames)

For every frame, the following data is provided:
- color: 			24-bit color image of the hand (cropped)
- joint_pos: 		3D joint positions relative to the middle MCP joint. The values are normalized such that the length between middle finger MCP and wrist is 1. 
					The positions are organized as a linear concatenation of the x,y,z-position of every joint (joint1_x, joint1_y, joint1_z, joint2_x, ...). 
					The order of the 21 joints is as follows: W, T0, T1, T2, T3, I0, I1, I2, I3, M0, M1, M2, M3, R0, R1, R2, R3, L0, L1, L2, L3.
					Please also see joints.png for a visual explanation.
- joint2D:			2D joint positions in u,v image coordinates. The positions are organized as a linear concatenation of the u,v-position of every joint (joint1_u, joint1_v, joint2_u, …).
- joint_pos_global:	3D joint positions in the coordinate system of the original camera (before cropping)
- crop_params:		cropping parameters that were used to generate the color image (256 x 256 pixels) from the original image (640 x 480 pixels), 
					specified as top left corner of the bounding box (u,v) and a scaling factor