Conservative Wasserstein Training for Pose Estimation

Abstract

This paper targets the task with discrete and periodic class labels ($e.g.,$ pose/orientation estimation) in the context of deep learning. The commonly used cross-entropy or regression loss is not well matched to this problem as they ignore the periodic nature of the labels and the class similarity, or assume labels are continuous value. We propose to incorporate inter-class correlations in a Wasserstein training framework by pre-defining ($i.e.,$ using arc length of a circle) or adaptively learning the ground metric. We extend the ground metric as a linear, convex or concave increasing function $w.r.t.$ arc length from an optimization perspective. We also propose to construct the conservative target labels which model the inlier and outlier noises using a wrapped unimodal-uniform mixture distribution. Unlike the one-hot setting, the conservative label makes the computation of Wasserstein distance more challenging. We systematically conclude the practical closed-form solution of Wasserstein distance for pose data with either one-hot or conservative target label. We evaluate our method on head, body, vehicle and 3D object pose benchmarks with exhaustive ablation studies. The Wasserstein loss obtaining superior performance over the current methods, especially using convex mapping function for ground metric, conservative label, and closed-form solution.

ICCV 2019

References in corpus (7)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Learning with a Wasserstein Loss
• Fast transport optimization for Monge costs on the circle
• Deep Verifier Networks: Verification of Deep Discriminative Models with Deep Generative Models
• Permutation-invariant Feature Restructuring for Correlation-aware Image Set-based Recognition
• Unsupervised Domain Adaptation via Calibrating Uncertainties
• Click Here: Human-Localized Keypoints as Guidance for Viewpoint Estimation

Cited by in corpus (9)

• Automated interpretation of congenital heart disease from multi-view echocardiograms
• Deep Verifier Networks: Verification of Deep Discriminative Models with Deep Generative Models
• AUTO3D: Novel view synthesis through unsupervisely learned variational viewpoint and global 3D representation
• Attention Control with Metric Learning Alignment for Image Set-based Recognition
• Reinforced Wasserstein Training for Severity-Aware Semantic Segmentation in Autonomous Driving
• Disentanglement for Discriminative Visual Recognition
• Energy-constrained Self-training for Unsupervised Domain Adaptation
• Importance-Aware Semantic Segmentation in Self-Driving with Discrete Wasserstein Training
• Embedding Semantic Hierarchy in Discrete Optimal Transport for Risk Minimization