Abstract:In response to the growth of digital photography and its many related applications, researchers have been actively investigating methods for providing automated aesthetical evaluation and classification of photographs. For computational networks to recognize aesthetic qualities, the learning algorithms must be trained using sample sets of characteristics that have known aesthetic values. Traditional methods for developing this training have required manual extraction of aesthetic features for use in the practice datasets. With abundant appearance of convolutional neural networks (CNN), the networks have learned features automatically and have acted as important tools for evaluation and classification. At the time of our research, several existing convolutional neural networks for photograph aesthetical classification only used shallow depth networks, which limit the improvement of performance. In addition, most methods have extracted only one patch as a training sample, such as a down-sized crop from each image. However, a single patch might not represent the entire image accurately, which could cause ambiguity during training. What's more, for existing datasets, the numbers of high quality images of each category are mostly too small to train deep CNN networks. To solve these problems, we introduce a novel photograph aesthetic classifier with a deep and wide CNN for fine-granularity aesthetical quality prediction. First, we download a large number of consumer photographic images from DPChallenge.com (a well-known online photography portal) to construct a dataset suitable for aesthetic quality assessment. Then, we zoom out the images into 256×256 by bilinear interpolation and crop 10 patches (Center+four Corners+Flipping). Once we have associated the set with the image's training labels, we feed the images with the bag of patches into the fine-tuned networks. Our proposed computational method is configured to classify the photographs into high and low aesthetic values. A training pattern specifying an output of (0, 1) indicates that the corresponding image belongs to the “low aesthetic quality” set. Likewise, a training pattern with an output of (1, 0) indicates that the corresponding image belongs to the “high aesthetic quality” set. Experimental results show that the accuracy of classification provided by our method is greater than 87.10%, which is noticeably better than the state-of-the-art methods. In addition, our experiments show that our results are fundamentally consistent with human visual perception and aesthetic judgments.

A System of Image Aesthetic Classification and Evaluation Using Cloud Computing.

A Comprehensive Survey on Computational Aesthetic Evaluation of Visual Art Images: Metrics and Challenges

An Efficient Image Aesthetic Analysis System Using Hadoop

Computational Aesthetics of APP Icons

Synthesized Computational Aesthetic Evaluation of Photos

Image Quality Assessment Based on Aesthetic Evaluation

Review for computational image aesthetics

CloudEye: A New Paradigm of Video Analysis System for Mobile Visual Scenarios

The Investigation of Cloud-Computing-based Image Mining Mechanism in Mobile Communication WEB on Android.

Photograph Aesthetical Evaluation and Classification with Deep Convolutional Neural Networks

A Biologically Inspired Automatic System for Media Quality Assessment

Social-sensed Image Aesthetics Assessment.

Image Data Model Optimization Method Based on Cloud Computing

Image Aesthetics Assessment Based on User Social Behavior.

Cloud Rendering Learning Platform Technology Research for Visual Analysis of Large Scale 3D Multimedia Data

Intelligent Image Detection System Based on Internet of Things and Cloud Computing

Aesthetic Quality Assessment Of Photos With Faces

Classification of Aesthetic Natural Scene Images Using Statistical and Semantic Features

IAACS: Image Aesthetic Assessment Through Color Composition and Space Formation

Implementation and Optimization of Image Acquisition with Smartphones in Computer Vision

Opinion-Unaware No-Reference Image Quality Assessment of Smartphone Camera Images Based on Aesthetics and Human Perception