Reproduce “Better, Stronger, Faster, and Broader: Structured All-Mask Prediction for MLLM-Based Segmentation” 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
Jiazhen Liu, Mingkuan Feng, Long Chen
STAMPlus extends structured all-mask prediction to multi-target segmentation by generating an autoregressive target list with explicit IDs, then predicting image-aligned mask labels for all listed targets together in one non-autoregressive pass. For 12 target categories, latency is 5.16 seconds versus 13.50 for repeated STAMP inference; Phase 1 remains autoregressive.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.