Search papers, labs, and topics across Lattice.
1
0
3
By predicting tracking models rather than image features, GOT-JEPA unlocks more robust object tracking, even when objects are heavily occluded or the environment is dynamic.
