Search papers, labs, and topics across Lattice.
Hangzhou Dianzi University
1
0
2
Calibrating text embeddings with visual prototypes boosts segmentation accuracy without any additional training, bridging the gap between generic concepts and specific visual instances.