BriefGPT.xyz
Aug, 2021
在线知识蒸馏用于高效姿态估计
Online Knowledge Distillation for Efficient Pose Estimation
HTML
PDF
Zheng Li, Jingwen Ye, Mingli Song, Ying Huang, Zhigeng Pan
TL;DR
本文提出了一种新型的在线知识蒸馏方法OKDHP,通过多分支网络和特征聚合单元以及像素级Kullback-Leibler散度来保证人体姿态估计的效率和准确性。实验结果表明,该方法在MPII和COCO数据集上表现良好。
Abstract
Existing state-of-the-art
human pose estimation
methods require heavy computational resources for accurate predictions. One promising technique to obtain an accurate yet lightweight pose estimator is
knowledge distillat
→