Robust-U1: MLLMs Recover Corrupted Images First
Robust-U1 trains an MLLM to reconstruct corrupted visual content, reaching 0.7398 overall on R-Bench versus 0.5770 for BAGEL and 0.5017 for Robust-R1.
Institution
Northwestern Polytechnical University
A Chinese engineering research university in Xi'an with active computer vision and intelligent systems work.