The objective assessment of aesthetic using improved convolution neural network

Shaoshuo Mu; Yanbing Jiang

doi:10.1117/12.2551979

31 January 2020 The objective assessment of aesthetic using improved convolution neural network

Shaoshuo Mu, Yanbing Jiang

Proceedings Volume 11427, Second Target Recognition and Artificial Intelligence Summit Forum; 114271W (2020) https://doi.org/10.1117/12.2551979
Event: Second Target Recognition and Artificial Intelligence Summit Forum, 2019, Changchun, China

Abstract

This paper proposes an improved neural network model based on CNN for evaluating the aesthetic quality of images. Firstly, we create three input sources from different angles of the target image and establish a three-channel parallel network model. Secondly, mlpconv layer is used to replace the traditional linear convolution layer to obtain more nonlinear abstract features in the network model based on CNN; Then, the combination of the global average pooling and full connection layers are used to replace the full connection layer in traditional CNN, and the three-channel features are merged. Finally, the EMD function is used as the loss function in the softmax layer. The output is probability density mass function from 1 to 10, and the mean and variance are used as objective qualitative score of picture quality. Experiments show that the proposed algorithm is feasible and effective, which solves the problem that the traditional method only obtains the binary classification of aesthetic. And this method gives the objective quantization score of the image. At the same time, the algorithm can get the evaluation value which is consistent with the actual situation in the real-time aerial experiment.

Citation Download Citation

Shaoshuo Mu and Yanbing Jiang "The objective assessment of aesthetic using improved convolution neural network", Proc. SPIE 11427, Second Target Recognition and Artificial Intelligence Summit Forum, 114271W (31 January 2020); https://doi.org/10.1117/12.2551979

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available