Self Adversarial Training for Human Pose Estimation

Abstract

This paper presents a deep learning based approach to the problem of human pose estimation. We employ generative adversarial networks as our learning paradigm in which we set up two stacked hourglass networks with the same architecture, one as the generator and the other as the discriminator. The generator is used as a human pose estimator after the training is done. The discriminator distinguishes ground-truth heatmaps from generated ones, and back-propagates the adversarial loss to the generator. This process enables the generator to learn plausible human body configurations and is shown to be useful for improving the prediction accuracy.

CVPR 2017 Workshop on Visual Understanding of Humans in Crowd Scene and the 1st Look Into Person (LIP) Challenge

References in corpus (4)

• BEGAN: Boundary Equilibrium Generative Adversarial Networks
• Human pose estimation via Convolutional Part Heatmap Regression
• Multi-Context Attention for Human Pose Estimation
• Adversarial PoseNet: A Structure-aware Convolutional Network for Human Pose Estimation

Cited by in corpus (19)

• Human Pose Regression by Combining Indirect Part Detection and Contextual Information
• Numerical Coordinate Regression with Convolutional Neural Networks
• Simple Baselines for Human Pose Estimation and Tracking
• Rethinking on Multi-Stage Networks for Human Pose Estimation
• Human Pose Estimation with Spatial Contextual Information
• Deep High-Resolution Representation Learning for Human Pose Estimation
• Look at Boundary: A Boundary-Aware Face Alignment Algorithm
• Multi-Scale Structure-Aware Network for Human Pose Estimation
• Integral Human Pose Regression
• Point-Set Anchors for Object Detection, Instance Segmentation and Pose Estimation
• Cross-domain Human Parsing via Adversarial Feature and Label Adaptation
• Dynamic Kernel Distillation for Efficient Pose Estimation in Videos
• Cross-Domain Adaptation for Animal Pose Estimation
• 3D Human Pose Estimation with 2D Marginal Heatmaps
• Global Context for Convolutional Pose Machines
• AMIL: Adversarial Multi Instance Learning for Human Pose Estimation
• TRB: A Novel Triplet Representation for Understanding 2D Human Body
• View Invariant 3D Human Pose Estimation
• MultiPoseNet: Fast Multi-Person Pose Estimation using Pose Residual Network