Reproduce “PSEE: Progressive Sensor Event Expansion for Point-Supervised Temporal Action Localization” 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
Jiaxi Yin, Ge Wang, Han Ding, Fei Wang
PSEE turns action timestamps into candidate segments for wearable-sensor localization. It combines learned class activation with sensor-transition evidence, adaptive ownership, and ranked boundary candidates, then trains standard detectors. Across four benchmarks, pseudo-boundary tIoU improves over adapted baselines, while SlimSTAD mAP averages 55.70% versus 35.95%. Simulated points mean annotation savings and real point performance remain unmeasured.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.