Reproduce “High Efficiency Image Compression for Large Visual-Language Models” 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
Binzhe Li, Shurun Wang, Shiqi Wang, Yan Ye
The authors present a variable-rate image compression system for sending images to large vision-language models. Visual-model tokens guide a pre-editing network trained with the codec using bitrate, image-distortion, and token losses. On OFA captioning on MSCOCO test, it saves 63.25% bitrate against VVC at matched CIDEr; evidence is limited to the tested models, datasets, and tasks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.