Human Pose Estimation with TensorFlow

Here you can find the implementation of the Human Body Pose Estimation algorithm, presented in the DeeperCut and ArtTrack papers:

Eldar Insafutdinov, Leonid Pishchulin, Bjoern Andres, Mykhaylo Andriluka and Bernt Schiele DeeperCut:  A Deeper, Stronger, and Faster Multi-Person Pose Estimation Model. In European Conference on Computer Vision (ECCV), 2016

Eldar Insafutdinov, Mykhaylo Andriluka, Leonid Pishchulin, Siyu Tang, Evgeny Levinkov, Bjoern Andres and Bernt Schiele ArtTrack: Articulated Multi-person Tracking in the Wild. In Conference on Computer Vision and Pattern Recognition (CVPR), 2017

For more information visit http://pose.mpi-inf.mpg.de

Prerequisites

The implementation is in Python 3 and TensorFlow. We recommended using conda to install the dependencies. First, create a Python 3.6 environment:

conda create -n py36 python=3.6
conda activate py36

Then, install basic dependencies with conda:

conda install numpy scikit-image pillow scipy pyyaml matplotlib cython

Install TensorFlow and remaining packages with pip:

pip install tensorflow-gpu easydict munkres

When running training or prediction scripts, please make sure to set the environment variableTF_CUDNN_USE_AUTOTUNE to 0 (see this ticketfor explanation).

If your machine has multiple GPUs, you can select which GPU you want to run on by setting the environment variable, eg. CUDA_VISIBLE_DEVICES=0.

Demo code

Single-Person (if there is only one person in the image)

# Download pre-trained model files
$ cd models/mpii
$ ./download_models.sh
$ cd -

# Run demo of single person pose estimation
$ TF_CUDNN_USE_AUTOTUNE=0 python3 demo/singleperson.py

Multiple People

# Compile dependencies
$ ./compile.sh

# Download pre-trained model files
$ cd models/coco
$ ./download_models.sh
$ cd -

# Run demo of multi person pose estimation
$ TF_CUDNN_USE_AUTOTUNE=0 python3 demo/demo_multiperson.py

Training models

Please follow these instructions

Citation

Please cite ArtTrack and DeeperCut in your publications if it helps your research:

@inproceedings{insafutdinov2017cvpr,
    title = {ArtTrack: Articulated Multi-person Tracking in the Wild},
    booktitle = {CVPR'17},
    url = {http://arxiv.org/abs/1612.01465},
    author = {Eldar Insafutdinov and Mykhaylo Andriluka and Leonid Pishchulin and Siyu Tang and Evgeny Levinkov and Bjoern Andres and Bernt Schiele}
}

@article{insafutdinov2016eccv,
    title = {DeeperCut: A Deeper, Stronger, and Faster Multi-Person Pose Estimation Model},
    booktitle = {ECCV'16},
    url = {http://arxiv.org/abs/1605.03170},
    author = {Eldar Insafutdinov and Leonid Pishchulin and Bjoern Andres and Mykhaylo Andriluka and Bernt Schiele}
}