Sorry, you need to enable JavaScript to visit this website.

facebooktwittermailshare

ICIP 2017 Poster Paper 3060

Abstract: 

Top-down attention plays an important role in guidance of human attention in real-world scenarios, but less efforts in computational modeling of visual attention has been put on it. Inspired by the mechanisms of top-down attention in human visual perception, we propose a multi-layer linear model of top-down attention to modulate bottom-up saliency maps actively. The first layer is a linear regression model which combines the bottom-up saliency maps on various visual features and objects. A contextual dependent upper layer is introduced to tune the parameters of the lower layer model adaptively. Finally, a mask of selection history is applied to the fused attention map to bias the attention selection towards the task related regions. Efficient learning algorithm with single-pass polynomial complexity is derived. We evaluate our model on a set of natural egocentric videos captured from a wearable glass in real-world environments. Our model outperforms the baseline and state-of-the-art bottom-up saliency models.

up
0 users have voted:

Paper Details

Authors:
Keng-Teck Ma, Liyuan Li, Peilun Dai, Joo-Hwee Lim, Chenyao Shen, Qi Zhao
Submitted On:
15 September 2017 - 1:33am
Short Link:
Type:
Poster
Event:
Presenter's Name:
Liyuan Li
Paper Code:
3060
Document Year:
2017
Cite

Document Files

ICIP2017_P3060.pdf

(189)

Subscribe

[1] Keng-Teck Ma, Liyuan Li, Peilun Dai, Joo-Hwee Lim, Chenyao Shen, Qi Zhao, "ICIP 2017 Poster Paper 3060", IEEE SigPort, 2017. [Online]. Available: http://sigport.org/2083. Accessed: Jul. 20, 2019.
@article{2083-17,
url = {http://sigport.org/2083},
author = {Keng-Teck Ma; Liyuan Li; Peilun Dai; Joo-Hwee Lim; Chenyao Shen; Qi Zhao },
publisher = {IEEE SigPort},
title = {ICIP 2017 Poster Paper 3060},
year = {2017} }
TY - EJOUR
T1 - ICIP 2017 Poster Paper 3060
AU - Keng-Teck Ma; Liyuan Li; Peilun Dai; Joo-Hwee Lim; Chenyao Shen; Qi Zhao
PY - 2017
PB - IEEE SigPort
UR - http://sigport.org/2083
ER -
Keng-Teck Ma, Liyuan Li, Peilun Dai, Joo-Hwee Lim, Chenyao Shen, Qi Zhao. (2017). ICIP 2017 Poster Paper 3060. IEEE SigPort. http://sigport.org/2083
Keng-Teck Ma, Liyuan Li, Peilun Dai, Joo-Hwee Lim, Chenyao Shen, Qi Zhao, 2017. ICIP 2017 Poster Paper 3060. Available at: http://sigport.org/2083.
Keng-Teck Ma, Liyuan Li, Peilun Dai, Joo-Hwee Lim, Chenyao Shen, Qi Zhao. (2017). "ICIP 2017 Poster Paper 3060." Web.
1. Keng-Teck Ma, Liyuan Li, Peilun Dai, Joo-Hwee Lim, Chenyao Shen, Qi Zhao. ICIP 2017 Poster Paper 3060 [Internet]. IEEE SigPort; 2017. Available from : http://sigport.org/2083