yoosan/video-understanding-dataset
A curated list of video understanding benchmark datasets including ActivityNet, Kinetics, Moments in Time, and UCF101.

Velocity · 7d
+0.2
★ / day
Trend
→steady
star history
This repository aggregates popular video understanding datasets used for computer vision research. It catalogs benchmark datasets for video classification and temporal action detection, including metrics, dataset sizes, class counts, and state-of-the-art performance references. The collection spans human action datasets (UCF101, HMDB51, Kinetics), activity recognition (ActivityNet), and temporal localization tasks (THUMOS).