AJAHR: Amputated Joint Aware 3D Human Mesh Recovery

2025-09-24

AI Summary

Existing 3D human mesh recovery methods rely on standard anatomical assumptions and thus fail to generalize to amputees, further hindered by the absence of dedicated datasets. To address this, we propose the first adaptive pose estimation framework tailored for amputees. Our method introduces: (1) an amputation-aware joint modeling mechanism that explicitly encodes residual limb geometry and biomechanical motion constraints; (2) a jointly optimized network for limb absence classification and 3D mesh regression; and (3) Amputee 3D (A3D), the first large-scale synthetic dataset of amputee poses, covering diverse amputation types and anatomical locations. Evaluated rigorously, our approach maintains state-of-the-art performance on non-amputee subjects while significantly outperforming prior methods on amputee cases—achieving new SOTA accuracy in 3D mesh recovery. This work establishes a foundational paradigm for inclusive human body modeling targeting underrepresented populations.

Technology Category

• Computer Vision: 3D Computer Vision
• Humans and AI: AI for Accessibility
• Intelligent Robots: Manipulation

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Existing human mesh recovery methods assume a standard human body structure, overlooking diverse anatomical conditions such as limb loss. This assumption introduces bias when applied to individuals with amputations - a limitation further exacerbated by the scarcity of suitable datasets. To address this gap, we propose Amputated Joint Aware 3D Human Mesh Recovery (AJAHR), which is an adaptive pose estimation framework that improves mesh reconstruction for individuals with limb loss. Our model integrates a body-part amputation classifier, jointly trained with the mesh recovery network, to detect potential amputations. We also introduce Amputee 3D (A3D), which is a synthetic dataset offering a wide range of amputee poses for robust training. While maintaining competitive performance on non-amputees, our approach achieves state-of-the-art results for amputated individuals. Additional materials can be found at the project webpage.

Problem

• Addressing bias in human mesh recovery for amputees
• Overcoming dataset scarcity for amputee pose estimation
• Improving 3D mesh reconstruction for limb loss conditions

Innovation

• Adaptive pose estimation for amputated individuals
• Body-part amputation classifier integrated with mesh recovery
• Synthetic Amputee 3D dataset for robust training