hAIr: A Hairstyle Recommender System Using Machine Learning Sark Xing Ward de Groot Abstract Technical University of Eindhoven Technical University of Eindhoven Choosing a new hairstyle can be a difficult, impactful
[email protected] [email protected] decision. Especially envisioning if a haircut would suit the individual is hard. With the analysis responses from Y izhou Liu Lara Leijtens facial recognition APIs and supervised machine Technical University of Eindhoven Technical University of Eindhoven learning, a relation between facial features and
[email protected] [email protected] hairstyle is ought to be found in this project, so that a hairstyle recommender system, called “hAIr”, can be created. The system recommends hairstyles that suit the individual’s characteristics. This is based on a neural network learning algorithm, which is trained with features, extracted from 1060 images of people, relating to 53 different hairstyles. The trained network reaches an accuracy of 28.10% when validated with images that were not used for training. This can be improved by trying different combinations of input variables, or using a different conversion for the values that were gained from the APIs. It is also possible that the APIs are not completely accurate. A third possibility for improvement would be to use a different learning algorithm, such as k-Nearest Neighbors or naive Bayes. Introduction about themselves, once they are presented with an Context option. To decide if they like the style or not can be Going to the hairdresser can come with difficult decision assessed by looking at another person that has that making.