Reproduce “GLD-Road:A global-local decoding road network extraction model for remote sensing images” 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
Ligao Deng, Yupeng Deng, Yu Meng, Jingbo Chen, Zhihao Xi, Diyou Liu, Qifeng Chu
A global-local decoding model, GLD-Road, extracts road networks from remote sensing imagery by combining global parallel and local iterative extraction strategies, yielding superior topological accuracy and efficiency compared to existing methods.
Key insights
- A hybrid global-local decoding approach can effectively reconcile the trade-off between extraction efficiency and topological accuracy in road network generation.
- Using a 36-dimensional directional descriptor provides a robust and fine-grained representation of road node connections, improving the accuracy of network topology.
- A denoising training strategy significantly stabilizes predictions and enhances topological accuracy, particularly in complex scenarios with dense road intersections.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.