Geometric and Semantic Coupling for Interaction Understanding in 3D Scenes
Paper • 2609.25247 • Published • 10
This repository contains the model presented in Geometric and Semantic Coupling for Interaction Understanding in 3D Scenes.
Project page: https://hyokong.github.io/segment-snap-page/
Code: https://github.com/HyoKong/Segment-Snap
Segment–Snap recovers movable parts, their motion, and the handles used to operate them from a 3D indoor scan. A door's surface constrains its possible motion, but its handle helps identify which side is hinged. In the other direction, a part provides context for finding and classifying small handles.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support