Traditional Bulgarian Costume Classification in a Tourist Guide application

Bulgaria is a culturally rich country which has been kept unexplored for years. Modern information and communication technologies can shrink the gap between tourists and its rich historical heritage. A digital tourist guide has been in development that offers its users a tailored experience and guides them through their journey in Bulgaria. Users are able to take a picture of a traditional Bulgarian costume and using various machine learning algorithms the tourist guide is able to recognize it and present interesting information about the photographed object. The following article gives an overview of the tourist guide architecture and image classification process. The authors used a convolutional neural network to classify the uploaded images of traditional Bulgarian costumes. The lack of dataset for training the network is solved by collecting the user-uploaded images, classifying them and using them for improving the classification algorithm automatically. Through this process a lifelong learning has been achieved.