Reproduce “CoGenRec: A Unified Generative Framework for Complementary Vision Reconstruction” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Yapeng Meng, Yuguo Chen, Xiangru Chen, Shaojun Xu, Xuezhen Wang, Lijian Wang, Zheyu Yang, Yihan Lin and 2 more

CoGenRec reconstructs fast scenes by combining dense RGB frames and high-speed colorless difference measurements with a pretrained video generator. The measurements guide generation and return through decoder skip connections to restore detail. On SportsSloMo-CVS, it improves interpolation PSNR by 3.89 dB over IGFNet across seven frames. This benchmark result does not establish pixel-accurate recovery in real scenes.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.