Search papers, labs, and topics across Lattice.
University of Southern California
2
0
6
Achieve up to 132x speedup in long-term conversational memory retrieval by intelligently skipping expensive dense retrieval steps, without sacrificing accuracy.
Revitalizing the latent edge-sensitivity of DINO with SAM's structural priors yields state-of-the-art open-vocabulary segmentation, especially in cluttered scenes.
