The Kinetics Human Action Video Dataset
Will Kay João Carreira Karen Simonyan Brian Zhang Chloe Hillier Sudheendra Vijayanarasimhan Fabio Viola Tim Green Trevor Back Paul Natsev Mustafa Suleyman Andrew Zisserman
Abstract
We describe the DeepMind Kinetics human action video dataset. The dataset contains 400 human action classes, with at least 400 video clips for each action. Each clip lasts around 10s and is taken from a different YouTube video. The actions are human focussed and cover a broad range of classes including human-object interactions such as playing instruments, as well as human-human interactions such as shaking hands. We describe the statistics of the dataset, how it was collected, and give some baseline performance figures for neural network architectures trained and tested for human action classification on this dataset. We also carry out a preliminary analysis of whether imbalance in the dataset leads to bias in the classifiers.
中文速览
大规模人类动作视频理解长期面临数据集太小、类别太少、同一视频反复切片导致多样性不足等瓶颈，为此 DeepMind 推出了 Kinetics 数据集：从 YouTube 抓取视频、用图像分类器自动定位动作片段、再交由 Amazon Mechanical Turk 人工核验，最终整理出 400 个人类动作类别、每类至少 400 条约 10 秒的短片，共超过 30 万条视频，且每条均来自不同的 YouTube 视频以保证多样性。测试结果表明，多种主流卷积网络架构在该数据集上的表现差异显著，说明它足以作为衡量不同模型能力的有效基准。Kinetics 规模远超此前通用的 HMDB-51 和 UCF-101 基准，能够从头训练深度网络，并像 ImageNet 之于图像领域那样，为视频理解的预训练与迁移学习提供坚实基础，有望推动视频领域新一代神经网络架构的发展。
原文 arXiv:1705.06950;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1705.06950v1