We present a method to analyze images taken from a passive egocentric wearable camera along with the contextual information, such as time and day of week, to learn and predict everyday activities of an individual. We collected a dataset of 40,103 egocentric images over a 6 month period with 19 activity classes and demonstrate the benefit of state-of-the-art deep learning techniques for learning and predicting daily activities. Classification is conducted using a Convolutional Neural Network (CNN) with a classification method we introduce called a late fusion ensemble. This late fusion ensemble incorporates relevant contextual information and increases our classification accuracy. Our technique achieves an overall accuracy of 83.07\% in predicting a person's activity across the 19 activity classes. We also demonstrate some promising results from two additional users by fine-tuning the classifier with one day of training data.
%0 Conference Paper
%1 CastroHicksonEtAl15ISWC
%A Castro, Daniel
%A Hickson, Steven
%A Bettadapura, Vinay
%A Thomaz, Edison
%A Abowd, Gregory
%A Christensen, Henrik
%A Essa, Irfan
%B Proceedings of the 2015 ACM International Symposium on Wearable Computers, Osaka, Japan
%C New York, NY, USA
%D 2015
%I ACM
%K 01614 acm paper embedded ai image action recognition learn zzz.vitra
%P 75--82
%R 10.1145/2802083.2808398
%T Predicting Daily Activities from Egocentric Images using Deep Learning
%X We present a method to analyze images taken from a passive egocentric wearable camera along with the contextual information, such as time and day of week, to learn and predict everyday activities of an individual. We collected a dataset of 40,103 egocentric images over a 6 month period with 19 activity classes and demonstrate the benefit of state-of-the-art deep learning techniques for learning and predicting daily activities. Classification is conducted using a Convolutional Neural Network (CNN) with a classification method we introduce called a late fusion ensemble. This late fusion ensemble incorporates relevant contextual information and increases our classification accuracy. Our technique achieves an overall accuracy of 83.07\% in predicting a person's activity across the 19 activity classes. We also demonstrate some promising results from two additional users by fine-tuning the classifier with one day of training data.
%@ 978-1-4503-3578-2
@inproceedings{CastroHicksonEtAl15ISWC,
abstract = {We present a method to analyze images taken from a passive egocentric wearable camera along with the contextual information, such as time and day of week, to learn and predict everyday activities of an individual. We collected a dataset of 40,103 egocentric images over a 6 month period with 19 activity classes and demonstrate the benefit of state-of-the-art deep learning techniques for learning and predicting daily activities. Classification is conducted using a Convolutional Neural Network (CNN) with a classification method we introduce called a late fusion ensemble. This late fusion ensemble incorporates relevant contextual information and increases our classification accuracy. Our technique achieves an overall accuracy of 83.07\% in predicting a person's activity across the 19 activity classes. We also demonstrate some promising results from two additional users by fine-tuning the classifier with one day of training data.},
added-at = {2016-05-22T14:58:53.000+0200},
address = {New York, NY, USA},
author = {Castro, Daniel and Hickson, Steven and Bettadapura, Vinay and Thomaz, Edison and Abowd, Gregory and Christensen, Henrik and Essa, Irfan},
biburl = {https://www.bibsonomy.org/bibtex/22d8ea8291e6ccca7766f2180a9504d3a/flint63},
booktitle = {Proceedings of the 2015 {ACM} International Symposium on Wearable Computers, Osaka, Japan},
doi = {10.1145/2802083.2808398},
file = {ACM Digital Library:2015/CastroHicksonEtAl15ISWC.pdf:PDF},
groups = {public},
interhash = {7962ffe3ed681b3418422b1052a2d879},
intrahash = {2d8ea8291e6ccca7766f2180a9504d3a},
isbn = {978-1-4503-3578-2},
keywords = {01614 acm paper embedded ai image action recognition learn zzz.vitra},
pages = {75--82},
publisher = {ACM},
timestamp = {2018-04-16T12:17:48.000+0200},
title = {Predicting Daily Activities from Egocentric Images using Deep Learning},
username = {flint63},
year = 2015
}