Reproduce “Learning to Cut by Watching Movies” 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

Alejandro Pardo, Fabian Caba Heilbron, Juan León Alcázar, Ali Thabet, Bernard Ghanem

The authors introduce Learning to Cut, which ranks joins between untrimmed shots using patterns learned from finished edits. It scores moments, then ranks pairs with contrastive learning and a cut-plausibility task. On edited-movie validation, it achieves 8.18% at R@1K, d=1, versus 2.89% for its single-stage version. In an unedited-footage study, editors preferred its cuts to baselines, but preferred professional cuts.

Summary by alphaXiv.

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