Learning Enhanced Resolution-wise features for Human Pose Estimation

TL;DR

This paper introduces a novel resolution-wise attention module and pyramid refinement technique to improve human pose estimation accuracy by learning better resolution-specific features, achieving state-of-the-art results on MS-COCO.

Contribution

The paper proposes a Resolution-wise Attention Module and Gradual Pyramid Refinement to enhance resolution-specific features for more accurate pose estimation.

Findings

Achieved 77.7 AP on MS-COCO val2017 set.

Achieved 77.0 AP on MS-COCO test-dev2017 set.

State-of-the-art performance without extra training data.

Abstract

Recently, multi-resolution networks (such as Hourglass, CPN, HRNet, etc.) have achieved significant performance on pose estimation by combining feature maps of various resolutions. In this paper, we propose a Resolution-wise Attention Module (RAM) and Gradual Pyramid Refinement (GPR), to learn enhanced resolution-wise feature maps for precise pose estimation. Specifically, RAM learns a group of weights to represent the different importance of feature maps across resolutions, and the GPR gradually merges every two feature maps from low to high resolutions to regress final human keypoint heatmaps. With the enhanced resolution-wise features learnt by CNN, we obtain more accurate human keypoint locations. The efficacies of our proposed methods are demonstrated on MS-COCO dataset, achieving state-of-the-art performance with average precision of 77.7 on COCO val2017 set and 77.0 on test-dev2017 set without using extra human keypoint training dataset.