Jointly Optimize Data Augmentation and Network Training: Adversarial Data Augmentation in Human Pose Estimation

Abstract

Random data augmentation is a critical technique to avoid overfitting in training deep neural network models. However, data augmentation and network training are usually treated as two isolated processes, limiting the effectiveness of network training. Why not jointly optimize the two? We propose adversarial data augmentation to address this limitation. The main idea is to design an augmentation network (generator) that competes against a target network (discriminator) by generating `hard' augmentation operations online. The augmentation network explores the weaknesses of the target network, while the latter learns from `hard' augmentations to achieve better performance. We also design a reward/penalty strategy for effective joint training. We demonstrate our approach on the problem of human pose estimation and carry out a comprehensive experimental analysis, showing that our method can significantly improve state-of-the-art models without additional data efforts.

CVPR 2018

References in corpus (9)

• DeepPose: Human Pose Estimation via Deep Neural Networks
• Generative Adversarial Text to Image Synthesis
• Human pose estimation via Convolutional Part Heatmap Regression
• Stacked Hourglass Networks for Human Pose Estimation
• Convolutional Pose Machines
• Semantic Jitter: Dense Supervision for Visual Comparisons via Synthetic Images
• Adversarial PoseNet: A Structure-aware Convolutional Network for Human Pose Estimation
• A Recurrent Encoder-Decoder Network for Sequential Face Alignment
• Human Pose Estimation using Deep Consensus Voting

Cited by in corpus (14)

• Deep High-Resolution Representation Learning for Human Pose Estimation
• A Unified Query-based Generative Model for Question Generation and Question Answering
• Adaptive Wing Loss for Robust Face Alignment via Heatmap Regression
• Joint Transmission Map Estimation and Dehazing using Deep Networks
• Bottom-Up Human Pose Estimation by Ranking Heatmap-Guided Adaptive Keypoint Estimates
• Quantized Densely Connected U-Nets for Efficient Landmark Localization
• Towards Robust RGB-D Human Mesh Recovery
• RePose: Learning Deep Kinematic Priors for Fast Human Pose Estimation
• CU-Net: Coupled U-Nets
• Learning to Forecast and Refine Residual Motion for Image-to-Video Generation
• Anti-Confusing: Region-Aware Network for Human Pose Estimation
• SSAH: Semi-supervised Adversarial Deep Hashing with Self-paced Hard Sample Generation
• Efficient Human Pose Estimation by Learning Deeply Aggregated Representations
• Synthesis of High-Quality Visible Faces from Polarimetric Thermal Faces using Generative Adversarial Networks