Architectural style classification of the Chinese traditional settlements using deep learning

Qing Han; Chao Yin

doi:10.1117/12.2667749

10 February 2023 Architectural style classification of the Chinese traditional settlements using deep learning

Qing Han, Chao Yin

Proceedings Volume 12552, International Conference on Geographic Information and Remote Sensing Technology (GIRST 2022); 125522X (2023) https://doi.org/10.1117/12.2667749
Event: International Conference on Geographic Information and Remote Sensing Technology (GIRST 2022), 2022, Kunming, China

Abstract

This paper investigates state-of-the-art deep learning techniques to achieve automatic architectural style classification of the Chinese traditional settlements. First, a new annotated dataset is built with six typical Chinese architectural styles, consisting of over 1000 web-crawled images and an original image collection of Chinese traditional settlements. Second, a state-of-the-art convolutional network named DenseNet is benchmarked on the new dataset to learn the effectiveness of the deep learning networks. Yet, the DenseNet network suffered server overfitting on the small-sized new dataset. Third, to overcome the common overfitting problem, a new deep learning framework named DenseNet-TL-Aug is developed by leveraging transfer learning (TL) and data augmentation (DA) techniques, e.g., AutoAugment. The experimental results demonstrate that the new developed framework achieves much better classification performance in classifying the Chinese traditional style images than the original DenseNet, significantly mitigating the overfitting problem. This study will contribute to automated landscape gene recognition as well as the design and development of traditional tourism.

Citation Download Citation

Qing Han and Chao Yin "Architectural style classification of the Chinese traditional settlements using deep learning", Proc. SPIE 12552, International Conference on Geographic Information and Remote Sensing Technology (GIRST 2022), 125522X (10 February 2023); https://doi.org/10.1117/12.2667749

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available