Reproduce “Dataset-Aware Cold-Start Active Learning for Annotation-Efficient 3D Medical Image 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
Rémi Hattat, Marine Beaumont, Charline Bertholdt, Gabriela Hossu, Olivier Morel, Bailiang Chen
CSCS selects unlabeled 3D medical-image volumes for one-shot segmentation annotation by combining typicality and reconstruction-based uncertainty. It sets their balance using rank correlation across the unlabeled pool and selects one high-scoring volume per embedding cluster. Across four benchmarks and three budgets, it ranked first in eight of twelve configurations. Results support segmentation scores, not reduced annotation time or universal advantage.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.