Introduction to Video Classification & Human Activity Recognition
The Dataset we are using is the UCF50 - Action Recognition Dataset.
UCF50 is an action recognition dataset which contains:
50 Action Categories consisting of realistic YouTube videos
25 Groups of Videos per Action Category
133 Average Videos per Action Category
199 Average Number of Frames per Video
320 Average Frames Width per Video
240 Average Frames Height per Video
26 Average Frames Per Seconds per Video
Source: https://colab.research.google.com/drive/1Yxsyc7qTr7KjTyz8qa6mJqb3TagthWK-?usp=sharing&authuser=2