Search papers, labs, and topics across Lattice.
1
0
3
Ditch slow, context-deprived image crops: DISC extracts high-fidelity CLIP embeddings directly from vision transformer layers for faster, more accurate open-set semantic mapping.
