A single flow-matching model for pairwise and multi-view registration, generalizing across scene scales and sensor modalities.
Composing short trajectory fragments into coherent long-horizon robot plans with a global energy and consistent boundary corrections.
Exposing language shortcuts in 3D reasoning benchmarks and introducing Real-3DQA to test whether models actually use visual geometry.
One generative formulation for point cloud registration and multi-part assembly, learning geometric symmetries without symmetry labels.
A synthetic driving dataset with continuously changing conditions for studying domain adaptation and multitask perception.
Testing whether deterministic uncertainty estimates remain reliable under distribution shifts and in dense prediction tasks.
Leveraging crowdsourced GPS data to improve and support road extraction from aerial imagery.
Combining satellite imagery with GPS data to improve road extraction quality
Stacked U-Nets, auxiliary supervision, and topology-aware post-processing for extracting connected roads from satellite imagery.