In this project, we implemented several machine learning models to predict the genres of movies using their overview descriptions and posters (since we want to predict genres for new movies, we didn't use features like ratings, which are not available for new movies). We further compared those models and made a machine learning pipeline for the best model. The best model is a concatenated model composed of an encoder for posters and an encoder for overview texts.
Our pipeline can be used to automatically tag the genres of movies on movie-hosting websites like Netflix and movie-database websites like TMDB.
Please view our final report here.
Please view our poster here.