When will you do what? - Anticipating Temporal Occurrences of Activities
Authors:
- Yazan Abu FarhaYFYazan Abu Farha
- Alexander RichardARAlexander Richard
- Juergen GallJGJuergen Gall
- 1 June 2018
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 5343-5352
Abstract
Analyzing human actions in videos has gained increased attention recently. While most works focus on classifying and labeling observed video frames or anticipating the very recent future, making long-term predictions over more than just a few seconds is a task with many practical applications that has not yet been addressed. In this paper, we propose two methods to predict a considerably large amount of future actions and their durations. Both, a CNN and an RNN are trained to learn future video labels based on previously seen content. We show that our methods generate accurate predictions of the future even for long videos with a huge amount of different actions and can even deal with noisy or erroneous input information.Keywords
All Related Versions
This publication has 23 references indexed in Scilit:
- Temporal Convolutional Networks for Action Segmentation and Detection2017
- Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset2017
- Learning Activity Progression in LSTMs for Activity Detection and Early Detection2016
- Temporal Action Localization in Untrimmed Videos via Multi-stage CNNs2016
- Anticipating Visual Representations from Unlabeled Video2016
- Structural-RNN: Deep Learning on Spatio-Temporal Graphs2016
- Anticipating Human Activities Using Object Affordances for Reactive Robotic ResponseIEEE Transactions on Pattern Analysis and Machine Intelligence, 2015
- The Language of Actions: Recovering the Syntax and Semantics of Goal-Directed Human Activities2014
- Max-Margin Early Event DetectorsInternational Journal of Computer Vision, 2013
- Learning and parsing video events with goal and intent predictionComputer Vision and Image Understanding, 2013