Movie Poster Genre Classification with Convolutional Neural Network
Marcellus Marcellus, Dyah Erny Herwindiati, Janson Hendryli · 2021
Not infrequently we do not have a clear plan of events to do, and chose to do an activity spontaneously. One of the activities that is usually done spontaneously is watching a movie in the cinema, and in this case, information about the movie we want to watch is most likely not known to the prospective viewer. Therefore, this paper is expected to help build a software to classify film genres based on image data input in the form of movie posters. By utilizing the intended software, potential viewers can make genre predictions to help choose the films they want to watch. The genre classification process is carried out using the Convolutional Neural Network. MobileNetv2 architecture was chosen due to the minimal computing cost of this architecture, and given computing power of hardware in form of a smart phone is relatively smaller than the computing power of hardware in the form of a computer. The final result produced by this model is the classification is in the form of generic film genres, which are divided into: romance, action, horror, fantasy and comedy.