A Short Note about Kinetics-600
This incremental work provides a larger and more diverse dataset for researchers in video action recognition.
The authors extended the DeepMind Kinetics human action dataset from 400 to 600 classes, with each class now containing at least 600 video clips, and introduced multilingual queries for data collection. They provided detailed statistics and baseline results using the I3D neural network architecture.
We describe an extension of the DeepMind Kinetics human action dataset from 400 classes, each with at least 400 video clips, to 600 classes, each with at least 600 video clips. In order to scale up the dataset we changed the data collection process so it uses multiple queries per class, with some of them in a language other than english -- portuguese. This paper details the changes between the two versions of the dataset and includes a comprehensive set of statistics of the new version as well as baseline results using the I3D neural network architecture. The paper is a companion to the release of the ground truth labels for the public test set.