/ddp

Deep Depth Pose model for 3D human pose estimation

Primary LanguageMATLABMIT LicenseMIT

3D Pose Estimation from Depth Maps using a Deep combination of Poses

This is support code for [1].

Visit the project web page for further information.



Quick start

This code has been tested on Ubuntu 16.04 with Matlab R2017a and MatConvNet-1.0-beta25
Inside Matlab, type the following commands:

run <matconvnet_root>/matlab/vl_setupnn.m  
cd <ddp_root>  
addpath(genpath(pwd))  
mj_testITOPmodel  

If everything worked fine, a new figure showing the estimated poses should be opened.
Alternatively, after the setup steps, you can run the following demo:

mj_testITOPmodel_frontal

Note that first of all you have to download the 'mat' data files as indicated in the folder 'Data'.

Code reference

How to load the original ITOP data?

An example of how to read the ITOP depth maps and skeletons can be found in:

Utils/mj_loadITOPdataFromH5

Qualitative results

Video with estimations on ITOP:

References

[1] Manuel J. Marin-Jimenez, Francisco J. Romero-Ramirez, Rafael Muñoz-Salinas, Rafael Medina-Carnicer
3D Pose Estimation from Depth Maps using a Deep combination of Poses
Journal of Visual Communication and Image Representation (in press), 2018

@Article{Marin18jvcir,
  author     = "Marin-Jimenez, M.J. and Romero-Ramirez, F.J. and Mu\~noz-Salinas, R. and Medina-Carnincer, R.",
  title      = "3D Pose Estimation from Depth Maps using a Deep combination of Poses",
  journal    = "Journal of Visual Communication and Image Representation",
  year       = "2018",
  doi  = "https://doi.org/10.1016/j.jvcir.2018.07.010",
  note = "In press"
}