Reproduce “Predicting the Needle in a Petabyte Scale Haystack: Open-Vocabulary Event Anticipation in Satellite Imagery” 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

Shreelekha Revankar, Mikhail Klassen, Creon Levit, Ash Hoover, Kavita Bala, Bharath Hariharan

The authors present a system that finds satellite changes and forecasts what locations may become. It represents before-and-after descriptions as directions in a frozen vision-language model’s embedding space, then uses a temporal model to predict future embeddings and completion time. On PlanetScope imagery with RS-SigLIP, it reaches max F1 of 0.935 and 0.245 mAP@10 9+ months before completion; completion is embedding-defined.

Summary by alphaXiv.

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