Reproduce “Beyond One-to-One: Rethinking the Referring 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

Yutao Hu, Qixiong Wang, Wenqi Shao, Enze Xie, Zhenguo Li, Jungong Han, Ping Luo

DMMI extends referring image segmentation to instructions that identify zero, one, or multiple regions. It trains a segmentation branch alongside an image-guided task that reconstructs erased entity phrases, using bidirectional feature alignment. On Ref-ZOM, DMMI reports oIoU 68.77, mIoU 68.21, and zero-target accuracy 87.02, exceeding LAVT; aggregate scores do not separately measure many-target performance.

Summary by alphaXiv.

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