DocumentCode :
2712489
Title :
Probabilistic learning of task-specific visual attention
Author :
Borji, Ali ; Sihite, Dicky N. ; Itti, Laurent
Author_Institution :
Dept. of Comput. Sci., Univ. of Southern California, Los Angeles, CA, USA
fYear :
2012
fDate :
16-21 June 2012
Firstpage :
470
Lastpage :
477
Abstract :
Despite a considerable amount of previous work on bottom-up saliency modeling for predicting human fixations over static and dynamic stimuli, few studies have thus far attempted to model top-down and task-driven influences of visual attention. Here, taking advantage of the sequential nature of real-world tasks, we propose a unified Bayesian approach for modeling task-driven visual attention. Several sources of information, including global context of a scene, previous attended locations, and previous motor actions, are integrated over time to predict the next attended location. Recording eye movements while subjects engage in 5 contemporary 2D and 3D video games, as modest counterparts of everyday tasks, we show that our approach is able to predict human attention and gaze better than the state-of-the-art, with a large margin (about 15% increase in prediction accuracy). The advantage of our approach is that it is automatic and applicable to arbitrary visual tasks.
Keywords :
Bayes methods; computer vision; eye; image motion analysis; learning (artificial intelligence); object tracking; 3D video games; bottom-up saliency modeling; computer vision; contemporary 2D video games; dynamic stimuli; eye movement recording; gaze prediction; human attention prediction; human fixation prediction; next attended location prediction; previous motor action; probabilistic learning; static stimuli; task-driven influence; task-specific visual attention; top-down influence; unified Bayesian approach; Bayesian methods; Computational modeling; Context; Games; Predictive models; Vectors; Visualization;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computer Vision and Pattern Recognition (CVPR), 2012 IEEE Conference on
Conference_Location :
Providence, RI
ISSN :
1063-6919
Print_ISBN :
978-1-4673-1226-4
Electronic_ISBN :
1063-6919
Type :
conf
DOI :
10.1109/CVPR.2012.6247710
Filename :
6247710
Link To Document :
بازگشت