A Short Note on the Kinetics-700 Human Action Dataset

Abstract

We describe an extension of the DeepMind Kinetics human action dataset from 600 classes to 700 classes, where for each class there are at least 600 video clips from different YouTube videos. This paper details the changes introduced for this new release of the dataset, and includes a comprehensive set of statistics as well as baseline results using the I3D neural network architecture.

added note about dangers of training on k700 and evaluating on k400/k600. arXiv admin note: text overlap with arXiv:1808.01340