Multiview-consistent Semi-supervised Learning For 3D Human Pose Estimation

Abstract

The best performing methods for 3D human pose estimation from monocular images require large amounts of in-the-wild 2D and controlled 3D pose annotated datasets which are costly and require sophisticated systems to acquire. To reduce this annotation dependency, we propose Multiview-Consistent Semi Supervised Learning (MCSS) framework that utilizes similarity in pose information from unannotated, uncalibrated but synchronized multi-view videos of human motions as additional weak supervision signal to guide 3D human pose regression. Our framework applies hard-negative mining based on temporal relations in multi-view videos to arrive at a multi-view consistent pose embedding. When jointly trained with limited 3D pose annotations, our approach improves the baseline by 25% and state-of-the-art by 8.7%, whilst using substantially smaller networks. Lastly, but importantly, we demonstrate the advantages of the learned embedding and establish view-invariant pose retrieval benchmarks on two popu

Stats

• citations54
• S2 citations—
• github stars0
• HF likes0
• heat score13.05
• arxiv keymitra2019multiview

Related papers

• View-invariant, Occlusion-robust Probabilistic Embedding For Human Pose (2020)8.82
• V-VIPE: Variational View Invariant Pose Embedding (2024)2.26
• Poseembroider: Towards A 3D, Visual, Semantic-aware Human Pose Representation (2024)6.34
• Multiview Image-based Localization (2025)0.00
• Self-supervised Modal And View Invariant Feature Learning (2020)0.00
• DISP6D: Disentangled Implicit Shape And Pose Learning For Scalable 6D Pose Estimation (2021)9.03
• When Regression Meets Manifold Learning For Object Recognition And Pose Estimation (2018)10.07
• Category-level Pose Retrieval With Contrastive Features Learnt With Occlusion Augmentation (2022)1.91