Search papers, labs, and topics across Lattice.
1
0
3
4
DINOv3's self-supervised features are surprisingly good at zero-shot in-context segmentation, beating fine-tuned models with a fraction of the parameters.
