HKUST-Aerial-Robotics/SG-Reg
[T-RO 2025] SG-Reg: Generalizable and Efficient Scene Graph Registration
Encodes scene graphs using vision foundation models (Grounded-SAM) and multi-modal semantic features—open-set semantics, spatial topology, and shape descriptors—enabling fully self-supervised training without ground-truth annotations. Generates both coarse node features and dense point features to support bandwidth-efficient multi-agent communication and coarse-to-fine localization in indoor SLAM systems. Evaluated on real RGB-D sequences (ScanNet, 3RScan) reconstructed via FM-Fusion semantic mapping.
134 stars. No commits in the last 6 months.
Stars
134
Forks
5
Language
Python
License
—
Category
Last pushed
Jul 20, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/computer-vision/HKUST-Aerial-Robotics/SG-Reg"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
hanruihua/ir-sim
A Python-based lightweight robot simulator designed for navigation, control, and learning
InhwanBae/NPSN
Official Code for "Non-Probability Sampling Network for Stochastic Human Trajectory Prediction...
InhwanBae/GPGraph
Official Code for "Learning Pedestrian Group Representations for Multi-modal Trajectory...
oskarnatan/end-to-end-driving
Implementation code for: End-to-end Autonomous Driving with Semantic Depth Cloud Mapping and...
InhwanBae/EigenTrajectory
Official Code for "EigenTrajectory: Low-Rank Descriptors for Multi-Modal Trajectory Forecasting...