Paper Info
Title
A Deep CNN with Focused Attention Objective for Integrated Object Recognition and Localization.
Abstract
We propose a novel deep convolutional neural network (CNN) architecture able to perform the integrated object recognition and localization tasks. We propose the Focused Attention (FA) objective that aims to optimize the network to learn features only from objects of interest while suppress those features from the background. As a result, the features extracted by the learned models can be used to accurately predict both the object category and the bounding box of the recognized object in the input image. Experimental results show that the proposed CNN architecture trained with the FA objective achieves better performances than original AlexNet in both the object localization and recognition tasks.
Year
Venue
Field
2016
PCM
Computer vision,Architecture,Pattern recognition,Computer science,Convolutional neural network,Artificial intelligence,Minimum bounding box,Cognitive neuroscience of visual object recognition
DocType
Citations 
PageRank 
Conference
0
0.34
References 
Authors
7
4
Name
Order
Citations
PageRank
Xiaoyu Tao1245.75
Chenyang Xu258523.07
yihong gong37300470.57
Jinjun Wang429115.86