VideoGraph: Recognizing Minutes-Long Human Activities in Videos

ICCV 2019, Workshop on Scene Graph Representation and Learning Many human activities take minutes to unfold. To represent them, related works opt for statistical pooling, which neglects the temporal structure. Others opt for convolutional methods, as CNN and Non-Local. While successful in learning t...

Full description

Saved in:
Bibliographic Details
Main Authors Hussein, Noureldien, Gavves, Efstratios, Smeulders, Arnold W. M
Format Journal Article
LanguageEnglish
Published 13.05.2019
Subjects
Computer Science - Computer Vision and Pattern Recognition
Online AccessGet full text

Cover

More Information
Summary:ICCV 2019, Workshop on Scene Graph Representation and Learning Many human activities take minutes to unfold. To represent them, related works opt for statistical pooling, which neglects the temporal structure. Others opt for convolutional methods, as CNN and Non-Local. While successful in learning temporal concepts, they are short of modeling minutes-long temporal dependencies. We propose VideoGraph, a method to achieve the best of two worlds: represent minutes-long human activities and learn their underlying temporal structure. VideoGraph learns a graph-based representation for human activities. The graph, its nodes and edges are learned entirely from video datasets, making VideoGraph applicable to problems without node-level annotation. The result is improvements over related works on benchmarks: Epic-Kitchen and Breakfast. Besides, we demonstrate that VideoGraph is able to learn the temporal structure of human activities in minutes-long videos.
DOI:10.48550/arxiv.1905.05143