Reproduce “HyperFlow: Class-Incremental Semantic Change Detection via Hyperbolic Flow” 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
Manon Béchaz, Léo Buecher, Emanuele Dalsasso, Ciprian Tomoiaga, Patrick Jayet, Marcin Detyniecki, Laetitia Chapel, Lucas Drumetz and 2 more
HyperFlow performs semantic change detection on aerial images while allowing land-cover categories to be refined in stages. A supervised segmentation model embeds pixels in hyperbolic space, and an unsupervised change-detection stage learns ordinary representation shifts and flags deviations. On HiChange, its base-task binary change F1 is 29.6 versus 25.1 for segmentation differencing; final 32-class SCS is 13.8, versus 9.1 and 10.0 for non-incremental alternatives.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.