Search papers, labs, and topics across Lattice.
University College London
2
0
3
5
Forget geometric LODs: tokenizing 3D shapes by semantic salience unlocks SOTA reconstruction and efficient autoregressive generation with 10x-1000x fewer tokens.
DINOv2's impressive unimodal performance doesn't translate to cross-modal understanding, but a simple training tweak can align embeddings across RGB, depth, and segmentation without sacrificing feature quality.
