Papers
6
Total Citations
196
H-Index
4
About
Gerard Pons-Moll is a leading researcher in computer vision and 3D human body modeling, with a particular focus on understanding and reconstructing how humans interact with objects and environments in three-dimensional space. His work sits at the intersection of computer graphics, robotics, and machine learning, targeting applications in virtual and mixed reality, human-robot collaboration, and behavioral analysis. Among his most influential contributions is the **BEHAVE** dataset and tracking framework (2022), which has garnered over 146 citations and established a foundational benchmark for studying human-object interactions in natural, unconstrained settings. This work addressed the critical challenge of generalizing across a vast diversity of objects and interaction types. Building on this, Pons-Moll developed visibility-aware tracking methods capable of reconstructing human-object interactions from a single RGB camera, pushing the boundaries of practical, low-cost 3D capture. His more recent work, including **Interaction Replica** (2024) and **PhySIC** (2025), extends these capabilities to dynamic scene understanding and physically plausible interaction reconstruction from single images — essential steps toward building accurate digital twins and intelligent robotic systems. Through rigorous dataset creation and innovative modeling approaches, Pons-Moll has meaningfully advanced the field's ability to digitize the complexity of human physical interaction with the world.
Research Focus
Key Achievements
Top Papers
- 1BEHAVE: Dataset and Method for Tracking Human Object Interactions146 citations · 2022
- 2Visibility Aware Human-Object Interaction Tracking from Single RGB Camera30 citations · 2023
- 3
- 4BEHAVE: Dataset and Method for Tracking Human Object Interactions6 citations · 2022
- 5
- 6