/pose-residual-network-pytorch

Code for the Pose Residual Network introduced in 'MultiPoseNet: Fast Multi-Person Pose Estimation using Pose Residual Network' paper https://arxiv.org/abs/1807.04067

Primary LanguagePythonOtherNOASSERTION

Pose Residual Network

This repository contains a PyTorch implementation of the Pose Residual Network (PRN) presented in our ECCV 2018 paper:

Muhammed Kocabas, Salih Karagoz, Emre Akbas. MultiPoseNet: Fast Multi-Person Pose Estimation using Pose Residual Network. In ECCV, 2018. arxiv

PRN is described in Section 3.2 of the paper.

Getting Started

We have tested our method on Coco Dataset

Prerequisites

python
pytorch
numpy
tqdm
pycocotools
progress
scikit-image

Installing

  1. Clone this repository git clone https://github.com/salihkaragoz/pose-residual-network-pytorch.git

  2. Install Pytorch

  3. pip install -r src/requirements.txt

  4. To download COCO dataset train2017 and val2017 annotations run: bash data/coco.sh. (data size: ~240Mb)

Training

python train.py

For more options look at opt.py

Testing

  1. Download pre-train model

  2. python test.py --test_cp=PathToPreTrainModel/PRN.pth.tar

Results

Results on COCO val2017 Ground Truth data.

 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets= 20 ] = 0.892
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets= 20 ] = 0.978
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets= 20 ] = 0.921
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets= 20 ] = 0.883
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets= 20 ] = 0.912
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 20 ] = 0.917
 Average Recall     (AR) @[ IoU=0.50      | area=   all | maxDets= 20 ] = 0.982
 Average Recall     (AR) @[ IoU=0.75      | area=   all | maxDets= 20 ] = 0.937
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets= 20 ] = 0.902
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets= 20 ] = 0.944

License

Citation

If you find this code useful for your research, please consider citing our paper:

@Inproceedings{kocabas18prn,
  Title          = {Multi{P}ose{N}et: Fast Multi-Person Pose Estimation using Pose Residual Network},
  Author         = {Kocabas, Muhammed and Karagoz, Salih and Akbas, Emre},
  Booktitle      = {European Conference on Computer Vision (ECCV)},
  Year           = {2018}
}