aihgf / deepcut-cnn

CNN architecture for articulated human pose estimation

DeeperCut

Here you can find the implementation of the CNN-based human body part detectors, used in the DeeperCut.

First of all, you should build Caffe and its Python bindings as described in the official documentation.

In order to run the demo of pose estimation execute the following:

# you will need to install python's click package, ex. by executing
$ pip install click

$ cd <caffe_dir>
$ export PYTHONPATH=`pwd`/python

# Download Caffe model files
$ cd models/deepercut
$ ./download_models.sh

# Run demo of single person pose estimation
$ cd ../../python/pose
$ python ./pose_demo.py image.png --out_name=prediction

Citation

Please cite Deep(er)Cut in your publications if it helps your research:

@article{insafutdinov2016deepercut,
    author = {Eldar Insafutdinov and Leonid Pishchulin and Bjoern Andres and Mykhaylo Andriluka and Bernt Schiele},
    url = {http://arxiv.org/abs/1605.03170}
    title = {DeeperCut: A Deeper, Stronger, and Faster Multi-Person Pose Estimation Model},
    year = {2016}
}

@inproceedings{pishchulin16cvpr,
    title = {DeepCut: Joint Subset Partition and Labeling for Multi Person Pose Estimation},
    booktitle = {CVPR'16},
    url = {},
    author = {Leonid Pishchulin and Eldar Insafutdinov and Siyu Tang and Bjoern Andres and Mykhaylo Andriluka and Peter Gehler and Bernt Schiele}
}