Reproduce “Video Summarization Using Fully Convolutional Sequence Networks” 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
Mrigank Rochan, Linwei Ye, Yang Wang
SUM-FCN adapts fully convolutional image-segmentation networks to label each video frame as selected or discarded, using temporal convolutions to combine context across frames and produce framewise predictions in parallel. In the augmented supervised setting on SumMe, it reports an F-score of 51.1, compared with 43.6 for the strongest listed competitor. TVSum results are comparable, and no speedup is measured.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.