This repository contains the implementation of paper "Pose-Aware Person Recognition" by Vijay Kumar, Anoop Namboodiri, Manohar Paluri, C V Jawahar published at CVPR17.
The implementation is based on Python Caffe.
- Download the datasets from the below links and place in data/ folder.
- PIPA (test): Link
- Hannah movie : Link
- IMDB : Link
- Soccer videos : Link
- Download the trained models and place in models/ folder.
- The models (baseline, pose-specific and pose estimator) are available at link
- To reproduce the results on PIPA test set, run run_PIPA.ipynb
- For recognition in movie scenario, run run_hannah.ipynb
- For recognition in soccer setting, run run_soccer.ipynb
- Change the data folder variable in these scripts according to your path.
- Replace the liblinear path to your correct liblinear installation directory.
If you use this code or data, please cite the following papers.
- Vijay Kumar, Anoop Namboodiri, Manohar Paluri, C V Jawahar, Pose-Aware Person Recognition, CVPR 2017.
- N. Zhang et al., Beyond Fronta Faces: Improving Person Recognition using Multiple Cues, CVPR 2014.
- Oh et al., Person Recognition in Personal Photo Collections, ICCV 2015.
- Li et al., A Multi-lvel Contextual Model for Person Recognition in Photo Albums, CVPR 2016.
- Ozerov et al., On Evaluating Face Tracks in Movies, ICIP 2013.