Reproduce “FMT: Lossless Image Compression in a Fixed Number of Parallel Steps” 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
Ammar Ali
FMT co-designs image reconstruction with entropy coding to decode pixel values in parallel while recovering 8-bit pixels exactly. Its schedule uses 72 sequential steps, bounded as image size grows although work per step increases. On Kodak, FMT 512×4 decoded 61× faster than ArIB-BPS on an A100 and produced files 0.2% smaller; this result is limited to tested codecs and settings.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.