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.

Photofeeler-D3: A Neural Network with Voter Modeling for Dating Photo Impression Prediction

Deep Facial Features for Personalized Attractiveness Prediction

Ethically aligned Deep Learning: Unbiased Facial Aesthetic Prediction

Computational Facial Attractiveness Prediction by Aesthetics-Aware Features

Some like it hot - visual guidance for preference prediction

Bio-Inspired Deep Attribute Learning Towards Facial Aesthetic Prediction

Putative Ratios of Facial Attractiveness in a Deep Neural Network

Beauty3DFaceNet: Deep Geometry and Texture Fusion for 3D Facial Attractiveness Prediction.

2.5D Facial Personality Prediction Based on Deep Learning

Predicting Social Perception from Faces: A Deep Learning Approach

3D facial attractiveness prediction based on deep feature fusion

Recent Developments on Deep Big Vision

Learning to see people like people

Deep Rank Learning For Facial Attractiveness

What is Beautiful is Not Always Good: Influence of Machine Learning-Derived Photo Attractiveness on Intention to Initiate Social Interactions in Mobile Dating Applications.

Data-driven Enhancement of Facial Attractiveness

Facial Attractiveness Prediction Using Psychologically Inspired Convolutional Neural Network (PI-CNN)

Photograph Aesthetical Evaluation and Classification with Deep Convolutional Neural Networks

An Universal Image Attractiveness Ranking Framework

A new humanlike facial attractiveness predictor with cascaded fine-tuning deep learning model

Automated Facial Trait Judgment and Election Outcome Prediction: Social Dimensions of Face