chenxuluo / orinet-demo Goto Github PK

View Code? Open in Web Editor NEW

25.0 5.0 4.0 388 KB

The testing code for OriNet

Home Page: http://bmvc2018.org/contents/papers/0289.pdf

Lua 56.95% MATLAB 43.05%

3d-human-pose human-pose-estimation human-pose keypoints

orinet-demo's People

Contributors

Stargazers

Watchers

Forkers

xiangdonglai issac8huxleg

orinet-demo's Issues

where is the h5 file for mpi dataset?

I am confused about put the h5 file in data/mpi/, is that file created by yourself?

Imagename

What's your imagename means? I need your h5 file for coding, and imagename is not string but array?

About the comparision in your paper

In your paper, you said,
"Note that [23] also use 2D data during training and camera parameters during
testing.."
[23] Xiao Sun, Jiaxiang Shang, Shuang Liang, and YichenWei. Compositional human pose
regression. ICCV, 2017.

But I can not find any description in [23] that indicates the use of camera parameters during
testing, could you help pointing out? Thanks.

How to determine `scale` value defined in `annot-test.h5` ?

I downloaded annot-test.h5 provided from README.md and reading the data using h5py ( one of Python library).

I've confirmed the annot-test.h5 stores scale value. On the other hand, original annotation provided from here does not have scale. How did you determine scale for each images ?