Search papers, labs, and topics across Lattice.
2
0
4
Forget monocular depth architectures: BuildMamba uses visual state-space models to achieve state-of-the-art building height estimation from satellite imagery, slashing RMSE by nearly a meter.
By fusing LiDAR and camera data with VLM-derived semantic priors, VLMFusionOcc3D achieves state-of-the-art 3D semantic occupancy prediction, especially in adverse weather conditions where other methods falter.
