Chronological Classification of Ancient Paintings of Mogao Grottoes Using Convolutional Neural Networks

Abstract

Mogao Grottoes are known as one of the three famous ancient Buddhist sculptural sites of China, which contain some of finest Buddhist paintings spanning a period of 1,000 years. Chronological classification of the ancient Buddhist paintings of Mogao Grottoes can help archaeologists and culture researchers to study the humanities, customs and economy of the corresponding eras. In this paper, this paper first perform an initial study on the effect of three state-of-the-art convolutional neural network based methods (AlexNet, VGG and ResNet) on chronological classification of paintings of Mogao Grottoes, and then propose a new network by replacing the last average pooling layer of ResNet-50 with a sequential layers. Experiments demonstrate that our method achieves higher classification accuracy than the three models and two existing chronological classification methods.