P_VggNet: A convolutional neural network (CNN) with pixel-based attention map

Kunhua Liu; Peisi Zhong; Yi Zheng; Kaige Yang; Mei Liu

doi:10.1371/journal.pone.0208497

P_VggNet: A convolutional neural network (CNN) with pixel-based attention map

PLoS One. 2018 Dec 12;13(12):e0208497. doi: 10.1371/journal.pone.0208497. eCollection 2018.

Authors

Kunhua Liu¹, Peisi Zhong¹, Yi Zheng¹, Kaige Yang¹, Mei Liu¹

Affiliation

¹ Advanced Manufacturing Technology Center, Shandong University of Science and Technology, Qingdao, Shandong province, China.

Abstract

Attention maps have been fused in the VggNet structure (EAC-Net) [1] and have shown significant improvement compared to that of the VggNet structure. However, in [1], E-Net was designed based on the facial action unit (AU) center and for facial AU detection only. Thus, for the use of attention maps in every image type, this paper proposed a new convolutional neural network (CNN) structure, P_VggNet, comprising the following parts: P_Net and VggNet with 16 layers (VggNet-16). The generation approach of P_Net was designed, and the P_VggNet structure was proposed. To prove the efficiency of P_VggNet, we designed two experiments, which indicated that P_VggNet could more efficiently extract image features than VggNet-16.

Publication types

Research Support, Non-U.S. Gov't

MeSH terms

Algorithms*
Attention
Humans
Image Enhancement / methods
Image Processing, Computer-Assisted / methods*
Neural Networks, Computer*
Pattern Recognition, Automated / methods*

Grants and funding

This work was supported by the National Natural Science Foundation of China (No. 51375282), the Key Research and Development Plan of Shandong Province (2018GGX106001), and Scientific and Technological Innovation Projects of Shandong Science and Technology School (Nos. SDKDYC180329 and SDKDYC180334). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.