Learning joint reconstruction of hands and manipulated objects

Estimating hand-object manipulations is essential for interpreting and imitating human actions. Previous work has made significant progress towards reconstruction of hand poses and object shapes in isolation. Yet, reconstructing hands and objects during manipulation is a more challenging task due to significant occlusions of both the hand and object. While presenting challenges, manipulations may also simplify the problem since the physics of contact restricts the space of valid hand-object configurations. For example, during manipulation, the hand and object should be in contact but not interpenetrate. In this work, we regularize the joint reconstruction of hands and objects with manipulation constraints. We present an end-to-end learnable model that exploits a novel contact loss that favors physically plausible hand-object constellations. Our approach improves grasp quality metrics over baselines, using RGB images as input. To train and evaluate the model, we also propose a new large-scale synthetic dataset, ObMan, with hand-object manipulations. We demonstrate the transferability of ObMan-trained models to real data.

PDF Abstract CVPR 2019 PDF CVPR 2019 Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
hand-object pose DexYCB HMO Average MPJPE (mm) 17.6 # 7
Procrustes-Aligned MPJPE - # 4
OCE - # 6
MCE - # 5
ADD-S - # 4
hand-object pose HO-3D v2 HMO Average MPJPE (mm) - # 6
ST-MPJPE 31.8 # 8
PA-MPJPE 11.0 # 6
OME - # 7
ADD-S - # 7

Results from Other Papers


Task Dataset Model Metric Name Metric Value Rank Source Paper Compare
3D Hand Pose Estimation FreiHAND Hasson et al. PA-MPVPE 13.2 # 28
PA-F@5mm 0.436 # 26
PA-F@15mm 0.908 # 26

Methods


No methods listed for this paper. Add relevant methods here