Reproduce “EM3M: An Electron Micrograph Dataset for Microstructural Segmentation and Generation” 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
Nan Wang, Zhiyi Xia, Yiming Li, Siyuan Zhang, Shi Tang, Zuxin Fan, Xi Fang, Haoyi Tao and 2 more
EM3M provides 5,091 electron micrographs with 2,985,660 instance masks, descriptions, a segmentation benchmark, and a text-conditioned image generator. UniAIMS adapts flow-based segmentation by removing Cellpose’s size model and adding scale jittering and training and implementation changes. On the full test set, it reaches 0.800 [email protected] versus Cellpose’s 0.711; its [email protected] falls to 0.340 at IoU 0.9, showing dense and difficult cases remain challenging.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.