Reproduce “CAST: Calibrating Correspondences for Training-Free Biomedical In-Context 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

Mingze Jiang, Jiashu Han, Haonan Chen, Zhiyi Li, Yilun Du, Kunzan Liu

CAST performs biomedical in-context segmentation using a frozen DINOv3 backbone, without task-specific training or gradient-based adaptation. It calibrates patch features under geometric changes, then reranks matches using neighborhood feature changes and target/non-target layout. With five source exemplars, it achieves an equal-weight average of 72.39% Dice across eight datasets, versus SegGPT’s 60.05%, and ranks first in six volumetric settings, but not every task.

Summary by alphaXiv.

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