Reproduce “Sufficient Frame Subsets for Video 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
Lahiru Wijayasingha
The authors introduce sufficient frame subsets to find compact video-frame sets that preserve a model’s prediction. The method iteratively removes and restores frames until a probability-drop threshold is met, imputing masked frames to keep input length fixed. On SSV2, the first subset scored −0.0987 versus 0.4630 for Facility-DINO, but evaluation covered only six random test examples per class.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.