Search papers, labs, and topics across Lattice.
Affiliation:
4
0
7
6
CT metal artifact reduction can be solved with orders of magnitude less data by cleverly prompting and fine-tuning a vision-language diffusion model, but only if you prevent the model from hallucinating waffles and petri dishes.
Image editing models already know how objects *move*—unlocking surprisingly good video interpolation with just a handful of examples.
Stop hand-labeling real-world videos: a meta-model can learn to verify tracker outputs and generate high-quality pseudo-labels for fine-tuning, achieving SOTA with less data.
