FAMOS: Feed-Forward 3D Articulation Modeling from Sparse Observations
FAMOS uses several sparse object views together to infer movable parts and their joint parameters.
The paper says the model takes an unordered set of partial point clouds, rather than depending on a single monocular observation. Its Multi-state Articulation Transformer alternates state-wise and global attention to combine articulation cues across views. The authors also add supervision for the observed motion span of each part and train with a procedural generator that produces self-annotated assets. They report consistent gains over feed-forward and optimization-based baselines on PartNet-Mobility, ACD, and ArtiCraft-10K. ArXiv · AI/CL/LG's note
The paper says the model takes an unordered set of partial point clouds, rather than depending on a single monocular observation. Its Multi-state Articulation Transformer alternates state-wise and global attention to combine articulation cues across views. The authors also add supervision for the observed motion span of each part and train with a procedural generator that produces self-annotated assets. They report consistent gains over feed-forward and optimization-based baselines on PartNet-Mobility, ACD, and ArtiCraft-10K. ArXiv · AI/CL/LG's note
score 5