Search papers, labs, and topics across Lattice.
1
0
3
1
Unlock geometric priors in video foundation models for zero-shot depth estimation with 163x less task-specific data, sidestepping the hallucination problems of generative methods and the data needs of discriminative ones.
