Reproduce “OpenEarth-OVS44: A Multimodal Remote Sensing Dataset for Open-Vocabulary 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
Yimin Wei, Hongruixuan Chen, Hao Zheng, Hao Liu, Jian Song, Yuting Zhu, Junshi Xia, Naoto Yokoya
OpenEarth-OVS44 is a benchmark for fine-grained multimodal remote-sensing open-vocabulary segmentation. It provides 290,587 expert-verified segments across 44 classes, and evaluates optical, synthetically cloudy optical-plus-SAR, and held-out geographic settings. On SegEarth-OV, merged Road and Water IoUs are 91.4 and 91.8, while subclass mIoUs are 6.4 and 12.9. This shows coarse labels can conceal subtype errors; it does not establish real-cloud performance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.