Reproduce “Bridging Low-Level Image Fusion and High-Level Semantic Understanding” 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
Guangyuan Dong, Lingdong Shen, Xudong Zhang, Haodong Jing, Haoyu Wang, Zihao Li, Zonghao Ying, Ziyu Song and 3 more
LANCE fuses multispectral and panchromatic satellite images using a structured scene record from closed scene, object, and layout vocabularies. A U-Net conditions fusion on the record; text–text and text–mask losses, including hard-negative queries, check fused outputs against it. On WorldView-III, an external detector scores 50.5 [email protected] versus RAMSF’s 44.1, and held-out localization scores 0.654 OV-F1 versus 0.582; LANCE remains closed-vocabulary.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.