Summarization of videos by analyzing affective state of the user through crowdsource
- Authors
- Singhal, Ashish; Kumar, Pradeep; Saini, Rajkumar; Roy, Partha Pratim; Dogra, Debi Prosad; Kim, Byung-Gyu
- Issue Date
- Dec-2018
- Publisher
- Elsevier B.V.
- Keywords
- Video summarization; Electroencephalogram (EEG); Crowdsourcing; Random Forest
- Citation
- COGNITIVE SYSTEMS RESEARCH, v.52, pp 917 - 930
- Pages
- 14
- Journal Title
- COGNITIVE SYSTEMS RESEARCH
- Volume
- 52
- Start Page
- 917
- End Page
- 930
- URI
- https://scholarworks.sookmyung.ac.kr/handle/2020.sw.sookmyung/4118
- DOI
- 10.1016/j.cogsys.2018.09.019
- ISSN
- 1389-0417
- Abstract
- Video summarization is one of the key techniques to access and manage a huge chunk of videos. Video summarization is used to extract the effective contents of a video sequence to generate a concise representation of its content. The involvement of crowdsourcing models in recent years is effectively used by researchers in E-commerce domain to increase the quality of the data contents. In this paper, we present a video summarization framework based on users emotion while they watch videos by analyzing cerebral activities through Electroencephalogram (EEG) signals. Three emotions, namely happy, sad and neutral have been extracted from the EEG signals. Video frames have been synchronized with EEG signals and tagged with various emotions. Finally, a crowdsourcing model has been used for effective summarization of the videos. The qualitative assessment of video summarization has been conducted with the help of user ratings using online Google Forms application. EEG signals of 28 users have been recorded while video streams of different emotions. An average accuracy of 83.93% has been recorded in emotion classification using crowdsourcing. Output summarized videos include dynamic video skims and the corresponding audio stream for better understanding. (C) 2018 Elsevier B.V. All rights reserved.
- Files in This Item
-
Go to Link
- Appears in
Collections - ICT융합공학부 > IT공학전공 > 1. Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.