RMPE: Regional Multi-person Pose Estimation

TL;DR

The paper introduces RMPE, a framework that improves multi-person pose estimation accuracy by handling bounding box inaccuracies through three novel components, significantly outperforming previous methods.

Contribution

It presents a new regional multi-person pose estimation framework with three innovative modules to address bounding box errors, enhancing accuracy in complex scenes.

Findings

Achieved 17% higher mAP on MPII dataset.

Effectively handles inaccurate bounding boxes and redundant detections.

Framework components improve robustness of pose estimation.

Abstract

Multi-person pose estimation in the wild is challenging. Although state-of-the-art human detectors have demonstrated good performance, small errors in localization and recognition are inevitable. These errors can cause failures for a single-person pose estimator (SPPE), especially for methods that solely depend on human detection results. In this paper, we propose a novel regional multi-person pose estimation (RMPE) framework to facilitate pose estimation in the presence of inaccurate human bounding boxes. Our framework consists of three components: Symmetric Spatial Transformer Network (SSTN), Parametric Pose Non-Maximum-Suppression (NMS), and Pose-Guided Proposals Generator (PGPG). Our method is able to handle inaccurate bounding boxes and redundant detections, allowing it to achieve a 17% increase in mAP over the state-of-the-art methods on the MPII (multi person) dataset.Our model and source codes are publicly available.