Developing an artificial intelligence framework for online destination image photos identification

Abstract

With the development of advanced technologies in computer science, such as deep learning and transfer learning, the tourism field is facing a more intelligent and automated future development environment. In this study, an artificial intelligence (AI) framework is developed to identify tourism photos without human interaction. Adopting online destination photos of Australia as a data source, the results show that the model combining a deep convolutional neural network and mixed transfer learning achieved the best image identification performance. This study identified 25 image classification categories covering all the tourism scenes to serve as a foundation for future tourism computer vision research. The results indicate that the AI photo identification framework is of great benefit for the understanding of projected destination images and enhancing tourism experiences. This study contributes to the existing literature by introducing an intelligent automation framework to big data research in the tourism field, as well as by advancing innovative methodologies of online destination image analysis. Practically, the proposed framework contributes to the marketing and management of smart destinations by offering a state-of-the-art data mining method. • Develops an artificial intelligence framework to identify tourism photos. • Identifies 25 image classification categories covering all tourism scenes. • Introduces an intelligent automation framework in tourism big data research. • Offers a state-of-art data mining method to the management of smart destinations.