Search papers, labs, and topics across Lattice.
Affiliation:
5
0
3
A novel reinforcement learning reward derived from positive pairs alone boosts keypoint learning stability and performance, even in low-texture environments.
TemporalLens reveals that many video detectors are blind to temporal context, while YOLO-3D demonstrates that preserving temporal depth can dramatically enhance detection performance.
Reliably measuring fish swimming patterns from video unlocks a non-invasive way to monitor their health and detect diseases early.
Finally, you can relight 3D Gaussian splats of outdoor scenes captured in the wild, opening the door to realistic scene manipulation and editing.
Achieve accurate crowd counting from a moving train camera by explicitly modeling physically plausible 3D motion, outperforming methods that ignore these constraints.
