Deep Multi-Patch Aggregation Network
for Image Style, Aesthetics, and Quality Estimation

Xin Lu
The Pennsylvania State University

Zhe Lin, Xiaohui Shen, Radomir Mech
Adobe Research

James Z. Wang
The Pennsylvania State University

Abstract:

This paper investigates problems of image style, aesthetics, and quality estimation, which require fine-grained details from high-resolution images, utilizing deep neural network training approach. Existing deep convolutional neural networks mostly extracted one patch such as a downsized crop from each image as a training example. However, one patch may not always well represent the entire image, which may cause ambiguity during training. We propose a deep multi-patch aggregation network training approach, which allows us to train models using multiple patches generated from one image. We achieve this by constructing multiple, shared columns in the neural network and feeding multiple patches to each of the columns. More importantly, we propose two novel network layers (statistics and sorting) to support aggregation of those patches. The proposed deep multi-patch aggregation network integrates shared feature learning and aggregation function learning into a unified framework. We demonstrate the effectiveness of the deep multi-patch aggregation network on the three problems, i.e., image style recognition, aesthetic quality categorization, and image quality estimation. Our models trained using the proposed networks significantly outperformed the state of the art in all three applications.


Full Paper
(PDF, 1.8MB)


Citation: Xin Lu, Zhe Lin, Xiaohui Shen, Radomir Mech and James Z. Wang, ``Deep Multi-Patch Aggregation Network for Image Style, Aesthetics, and Quality Estimation,'' International Conference on Computer Vision, pp. 990-998, Santiago, Chile, IEEE, 2015.

© 2015 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.

Last Modified: September 25, 2015
© 2015