This is the pytorch implementation for OKDHP (ICCV 2021). This repository is extended based on FPD.
Links: https://arxiv.org/abs/2108.02092
In this work, we investigate a novel Online Knowledge Distillation framework by distilling Human Pose structure knowledge in a one-stage manner to guarantee the distillation efficiency, termed OKDHP. Specifically, OKDHP trains a single multi-branch network and acquires the predicted heatmaps from each, which are then assembled by a Feature Aggregation Unit (FAU) as the target heatmaps to teach each branch in reverse. Instead of simply averaging the heatmaps, FAU which consists of multiple parallel transformations with different receptive fields, leverages the multi-scale information, thus obtains target heatmaps with higher-quality. Specifically, the pixelwise Kullback-Leibler (KL) divergence is utilized to minimize the discrepancy between the target heatmaps and the predicted ones, which enables the student network to learn the implicit keypoint relationship. Besides, an unbalanced OKDHP scheme is introduced to customize the student networks with different compression rates. The effectiveness of our approach is demonstrated by extensive experiments on two common benchmark datasets, MPII and COCO.
In this code, you can reproduce the experiment results in the paper, including MPII and COCO.
- Running 4-Stack OKDHP with three branches on MPII dataset.
(Running based on one NVIDIA RTX 3090 GPU)
sh start.sh
@inproceedings{li2021online,
title={Online Knowledge Distillation for Efficient Pose Estimation},
author={Li, Zheng and Ye, Jingwen and Song, Mingli and Huang, Ying and Pan, Zhigeng},
booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
pages={11740--11750},
year={2021}
}