Loading…
Detecting user attention to video segments using interval EEG features
•A method of detecting the top 20% of viewer attention to video segments is proposed.•This is the first study of detecting viewer attention during video viewing.•All subject-independent models unbiased to specific genres are evaluated.•The all-14-channel, single-channel, and selected multi-channel m...
Saved in:
| Published in: | Expert systems with applications 2019-01, Vol.115, p.578-592 |
|---|---|
| Main Authors: | , , , |
| Format: | Article |
| Language: | English |
| Subjects: | |
| Citations: | Items that this one cites Items that cite this one |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | •A method of detecting the top 20% of viewer attention to video segments is proposed.•This is the first study of detecting viewer attention during video viewing.•All subject-independent models unbiased to specific genres are evaluated.•The all-14-channel, single-channel, and selected multi-channel models are included.•The interval band ratio features are the most suitable for all the types of models.
To manage voluminous viewed videos, which US adults watch at a rate of more than five hours per day on average, an automatic method of detecting highly attended video segments during video viewing is required to access them for fine-grained sharing and rewatching. Most electroencephalography (EEG)-based studies of user state analysis have addressed the recognition of attention-related states in a specific task condition, such as drowsiness during driving, attention during learning, and mental fatigue during task execution. In contrast to attention in a specific task condition, both inattention and normal attention are meaningless to viewers in terms of managing viewed videos, while detecting high attention paid to video segments would make a valuable contribution to an automatic management system of viewed videos based on viewer attention. To the best of our knowledge, this is the first EEG-based study of detecting viewer attention paid to video segments. This study describes how to collect video-induced EEG and attention data for video segments from viewers without bias to specific genres and how to construct a subject-independent detection model for the top 20% of viewer attention. The attention detection model using the proposed interval EEG features from 14 channels achieved the best average F1 score of 39.79% with an average accuracy of 52.96%. Additionally, this paper proposes a channel-based feature selection method that considers both the performances of single-channel models and their physical locations for investigating the group of channels relevant to attention detection. The attention detection models using the interval EEG features from all four or some of the channels located in the fronto-central, parietal, temporal, and occipital lobes of the left hemisphere achieved the best F1 score of 39.60% with an average accuracy of 48.70%. It is shown that these models achieve better performance than models using the features from all four or some of their symmetric channels in the right hemisphere and models using the features from six channels located |
|---|---|
| ISSN: | 0957-4174 1873-6793 |
| DOI: | 10.1016/j.eswa.2018.08.016 |