FMT: lossless image compression with fixed-step parallel GPU decoding in 72 steps
LEAD-ITMO · hf · 2026-11-18
LEAD-ITMO introduces FMT (Fixed-step Multiscale Tree), a lossless image compression format with fixed-time decoding.
- A coarse-to-fine tree structure organizes the image; a 512×4 neural decoder recovers it in a fixed 72 steps via parallel GPU prediction, making decode latency content-independent.
- Validated on Kodak; feature maps and decoder checkpoints are released, pitched at feeding training data faster.
More from Research
- Voyager 1 Reaches One Light-Day From Earth, Signals Now Take 24 Hours — snowclipsed · 2026-11-18
- TCS Hiring Statement Urges Quality Over Quantity as AI Fuels Paper Arms Race — fortnow · 2026-11-18
- New paper: a sociocultural approach to linguistic creativity — abenitezburraco · 2026-11-18
- 27B Model Rewrites Its Own Weights Without Training or Gradients, Paper Claims — udmrzn · 2026-11-18
- From 'impossible' to a working Bluetooth beacon: open-source EDA chip design is real — IgorCarron · 2026-11-18
- AURA 2.0: a streaming video LLM that speaks first, calls tools, and grounds visually — Shuhan Li · 2026-11-18