Parsing Videos of Actions with Segmental Grammars
Authors:
- Hamed PirsiavashHPHamed Pirsiavash
- Deva RamananDRDeva Ramanan
- 1 June 2014
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 10636919,p. 612-619
Abstract
Real-world videos of human activities exhibit temporal structure at various scales, long videos are typically composed out of multiple action instances, where each instance is itself composed of sub-actions with variable durations and orderings. Temporal grammars can presumably model such hierarchical structure, but are computationally difficult to apply for long video streams. We describe simple grammars that capture hierarchical temporal structure while admitting inference with a finite-state-machine. This makes parsing linear time, constant storage, and naturally online. We train grammar parameters using a latent structural SVM, where latent subactions are learned automatically. We illustrate the effectiveness of our approach over common baselines on a new half-million frame dataset of continuous YouTube videos.Keywords
This publication has 25 references indexed in Scilit:
- Detecting activities of daily living in first-person camera views2012
- A survey on vision-based human action recognitionImage and Vision Computing, 2010
- Modeling Temporal Structure of Decomposable Motion Segments for Activity Classification2010
- Learning structural SVMs with latent variables2009
- Action MACH a spatio-temporal Maximum Average Correlation Height filter for action recognition2008
- Retrieving actions in movies2007
- Event Detection in Crowded Videos2007
- Layered representations for learning and inferring office activity from multiple sensory channelsComputer Vision and Image Understanding, 2004
- Parametric hidden Markov models for gesture recognitionIEEE Transactions on Pattern Analysis and Machine Intelligence, 1999
- Detection and Recognition of Periodic, Nonrigid MotionInternational Journal of Computer Vision, 1997