Multi-task based Image Aesthetics Quality Evaluation

被引：1

作者：

Jiang, Min ^{[1
,2
]}

Chen, Zhe ^{[1
,2
]}

Jiang, Jiajun ^{[1
,2
]}

Liu, Xiaoming ^{[1
,2
]}

Hu, Wei ^{[1
,2
]}

机构：

[1] Wuhan Univ Sci & Technol, Dept Comp Sci & Engn, Wuhan 430065, Hubei, Peoples R China

[2] Wuhan Univ Sci & Technol, Hubei Prov Key Lab Intelligent Informat Proc & Re, Wuhan 430065, Hubei, Peoples R China

来源：

2021 IEEE INTERNATIONAL CONFERENCE ON SYSTEMS, MAN, AND CYBERNETICS (SMC) | 2021年

基金：

中国国家自然科学基金;

关键词：

Deep Learning; Multi-task; Image Aesthetics quality evaluation;

D O I：

10.1109/SMC52423.2021.9659217

中图分类号：

TP3 [计算技术、计算机技术];

学科分类号：

0812 ;

摘要：

Image aesthetics quality evaluation, which allows computers to judge "beauty and ugliness", is widely used in fields of image recommendation and image editing etc.. Most aesthetic evaluation methods can only output one type of evaluation result, thus, the scope of their application scenarios is limited. To solve this problem, we proposed a multi-task based aesthetic evaluation system, which can output the image style label and three forms of the aesthetic evaluation results for a image. The training procedure is divided into two stages to realize the aesthetic evaluation tasks from coarse to fine. Experiments on AVA dataset show that it can accomplish an efficient and comprehensive image aesthetic quality evaluation.

引用

页码：2755 / 2760

页数：6

共 24 条

[1] Predicting Image Aesthetics with Deep Learning [J].