Search papers, labs, and topics across Lattice.
2
0
3
TOLiD achieves superior transfer learning in LiDAR representation by effectively bridging the modality and architecture gaps between vision and 3D data.
Monocular depth priors, previously too unreliable for 3D Gaussian Splatting, can now be effectively harnessed to boost rendering quality by focusing regularization on ill-posed geometry.
