Search papers, labs, and topics across Lattice.
1
0
3
RIS models struggle with motion-based queries, but a new data augmentation and contrastive learning approach closes the gap without sacrificing performance on appearance-based descriptions.
