Gaze movement inference for user adapted image annotation and retrieval

Authors:
S. Navid H. Haji Mirza;Michael Proulx;Ebroul Izquierdo
Affiliations:
Queen Mary University of London, London, United Kingdom;Queen Mary University of London, London, United Kingdom;Queen Mary University of London, London, United Kingdom
Venue:
SBNMA '11 Proceedings of the 2011 ACM workshop on Social and behavioural networked media access
Year:
2011

Citing 9
Cited 0

Real-time eye detection and tracking under various light conditions

ETRA '02 Proceedings of the 2002 symposium on Eye tracking research & applications
ViewPointer: lightweight calibration-free eye tracking for ubiquitous handsfree deixis

Proceedings of the 18th annual ACM symposium on User interface software and technology
Can relevance of images be inferred from eye movements?

MIR '08 Proceedings of the 1st ACM international conference on Multimedia information retrieval
Extracting Takagi-Sugeno Fuzzy Rules with Interpretable Submodels via Regularization of Linguistic Modifiers

IEEE Transactions on Knowledge and Data Engineering
GaZIR: gaze-based zooming interface for image retrieval

Proceedings of the 2009 international conference on Multimodal interfaces
Normalized mutual information feature selection

IEEE Transactions on Neural Networks
Image ranking with implicit feedback from eye movements

Proceedings of the 2010 Symposium on Eye-Tracking Research & Applications
Toward general type-2 fuzzy logic systems based on zSlices

IEEE Transactions on Fuzzy Systems
Enhanced Fuzzy System Models With Improved Fuzzy Clustering Algorithm

IEEE Transactions on Fuzzy Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

In media personalisation the media provider needs to receive feedbacks from its users to adapt media contents used for interaction. At the current stage this feedback is limited to mouse clicks and keyboard entries. This report explores the possible solutions to include the gaze movements of a user as a form of feedback for media personalisation and adaptation. Features are extracted from the gaze trajectory of users while they are searching in an image database for a Target Concept(TC). These features are used to measure a user's visual attention to every image appeared on the screen called user interest level(UIL). Because the reaction of different people to the same content are different, for every new user a new adapted processing interface is developed automatically. In average our interface could detect 10% of the images belonging to the TC class with no error and it could identify 40% of them with only 20% error. We show in this paper that the gaze movement is a reliable feedback to be used for measuring one's interest to images which help to personalise image annotation and retrieval.