UCF-101-24

The UCF101-24 dataset contains a collection of videos categorized into 24 action classes and is used to evaluate few-shot common action localization in time and space.

Papers using UCF-101-24 (3)

Dance With Flow: Two-in-one Stream Action Detection2019 · 95 citesModeling Spatio-Temporal Human Track Structure for Action Localization2018 · 4 citesFew-Shot Transformation of Common Actions into Time and Space2021