Abstract:2D human pose estimation has essential applications in traffic prediction and human-computer interaction. We propose a pose refinement network for refining human pose features to improve human pose detection accuracy. We propose our approach to two critical problems in pose estimation, occlusion and background interference, for which the model often needs local information for inference. In contrast, for the background interference problem, the model needs more non-local semantic information to guide the model to make semantic judgments. Specifically, we propose Joint Constraint-Based Human Pose Refinement Networks, which explicitly and implicitly model the relationship between critical points through graph neural networks and self-attention mechanisms to capture local and non-local information of the pose features and refine the keypoints responses. In order to avoid mutual conflicts between explicit and implicit modeling, this paper chooses a parallel structure. It selects an appropriate feature fusion strategy to fuse parallel branch features to improve model robustness. Meanwhile, to make the model pay more attention to the hard-to-identify vital points, this paper adopts the focal loss function to optimize the model and improve the data’s long-tailed distribution. We validate our work on the publicly available dataset MPII with the MSCOCO dataset. The results show that our proposed method has the effect of refining the keypoints response, with a PCKh@0.5 of 92.51% on the official MPII validation dataset, and achieved a result of 78% AP on the official MSCOCO validation dataset, which is comparable to the performance of existing state-of-the-art models.

Lightweight Whole-Body Human Pose Estimation With Two-Stage Refinement Training Strategy

Effective Whole-body Pose Estimation with Two-stages Distillation

Single-Network Whole-Body Pose Estimation

Learning to Refine Human Pose Estimation

Two-Stage Representation Refinement Based on Convex Combination for 3D Human Poses Estimation

Lightweight and Effective Human Pose Estimation Model Based on Multi-Angle Knowledge Distillation

Lightweight and efficient human pose estimation with enhanced priori skeleton structure

Lightweight Human Pose Estimation Based on Adaptive Feature Sensing

A Lightweight Top-Down Multi-Person Pose Estimation Method Based on Symmetric Transformation and Global Matching

Knowledge Distillation for Lightweight 2D Single-Person Pose Estimation

An improved lightweight high-resolution network based on multi-dimensional weighting for human pose estimation

Simple and Lightweight Human Pose Estimation

Lightweight Non-local High-Resolution Networks for Human Pose Estimation.

Human Pose Estimation Using Exemplars and Part Based Refinement

PLPose: A Bottom-up Lightweight Pose Estimation Detection Model

OpenPose : Whole-Body Pose Estimation

Efficient Pose Estimation Via a Lightweight Single-Branch Pose Distillation Network

Bidirectional Optimization Coupled Lightweight Networks for Efficient and Robust Multi-Person 2D Pose Estimation

Efficient and Lightweight Human Pose Estimation Algorithm

JCN: Joint Constraint-Based Human Pose Refinement Networks