Image Captioning for Thai Cultures

Abstract

Before each trip, tourists generally gather information or photos from different places. This work aims at providing additional information about touristic sites in Thailand via automatic image captioning. Image captioning is the process of generating a textual description for given images. In recent years, the development of Artificial Intelligence in combining image processing and natural language processing has gained attention worldwide. Image captioning can be regarded as a sequence-to-sequence modeling problem, as it converts images, which are considered a sequence of pixels, to a sequence of words. This work proposed a finetuned model that combined CNNs and LSTM to generate the image description. In the experiment part, we use BLEU to evaluate the model.