Reproduce “ConsensusTAS: Self-Supervised Temporal Action Segmentation for Long-Horizon Construction Videos” 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
Xiaoshan Zhou, Yafei Sun
ConsensusTAS proposes anonymous temporal segmentation to discover phases in unlabeled video. It generates candidate boundaries at multiple scales, scores them using agreement and video evidence, and decodes a density of boundaries without action labels. On Assembly101, it scores 33.50 ± 0.24 F1@50 versus 26.21 ± 0.18 for ASESM, but trails at F1@10; one bricklaying clip offers tentative human-reviewed evidence.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.