Visual aesthetic analysis using deep neural network: model and techniques to increase accuracy without transfer learning
This work addresses image aesthetic assessment for computer vision applications, but it is incremental as it uses standard techniques without major innovations.
The authors tackled visual aesthetic analysis by training a deep CNN from scratch without transfer learning, achieving 78.7% accuracy on the AVA2 dataset, close to the best models at 85.6%, and improved to 81.48% with more data.
We train a deep Convolutional Neural Network (CNN) from scratch for visual aesthetic analysis in images and discuss techniques we adopt to improve the accuracy. We avoid the prevalent best transfer learning approaches of using pretrained weights to perform the task and train a model from scratch to get accuracy of 78.7% on AVA2 Dataset close to the best models available (85.6%). We further show that accuracy increases to 81.48% on increasing the training set by incremental 10 percentile of entire AVA dataset showing our algorithm gets better with more data.