Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

A Short Note about Kinetics-600

About

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.

Joao Carreira, Eric Noland, Andras Banki-Horvath, Chloe Hillier, Andrew Zisserman• 2018

Related benchmarks

TaskDatasetResultRank
Action RecognitionKinetics-400 1.0 (val)
Top-1 Accuracy72.1
110
Video ClassificationKinetics-600
Top-1 Accuracy71.6
84
Video ClassificationKinetics-600 (val)
Accuracy71.9
84
Action RecognitionKinetics-600 (val)
Top-1 Acc71.9
68
Action DetectionAVA v2.1 (val)
mAP21.9
48
Action DetectionAVA v2.1 (test)
mAP21
3
Showing 6 of 6 rows

Other info

Code

Follow for update