CASE STUDY: THE CLASSIFICATION OF THE ROOMS IN HOLIDAY HOMES WITH DEEP LEARNING

Abstract

From reservation to the accommodation process, the effects of technology are increasing day by day in the field of tourism. Online booking platforms, virtual support assistants, mobile applications, and artificial intelligence tools can be given as examples. In the focus on artificial intelligence for tourism, different tools can be presented as examples, especially price analysis regression/recommendations, room, house & amenity classifications from images, and occupancy estimations. Our case study consists of two different steps. First, a dataset was created from a German-based tourism reservation company. In the second step, 5 different deep learning models were trained to compare the accuracy and loss with the dataset. We trained ResNet, DenseNet, VGGNet, Inception v3, and NASNet models. The following accuracies were observed based on 20 epochs of training; ResNet 97.4%, DenseNet 98.69%, VGGNet 97.31%, Inception v3 97.33%, and NASNet 97.21%.