Movie Genre Classification with Deep Neural Network using Poster Images

Abstract

The utility of the deep learning model on poster images to classify their corresponding movie genre is an incredible challenge in computer vision. The objective of this study is to make a computerized framework that works with pictures, both low-level and high-level highlights (such as object detection and form) to classify movie genres. Categorizing motion pictures makes it easier for the watcher to choose a movie they would like to watch. The method of categorizing movies is known as a genre. Many people started to work with low-level features (colour, edge) utilizing deep neural networks for movie genre classification. In this work, a convolutional neural network has been applied to classify the movie genre using their corresponding poster images. At first, this study extracted features from the movie's poster images and then classified the movie genres. A multilayered convolutional neural network is outlined which is trained over a large number of poster images. This study reached an accuracy of 91.15%, f1 score of 0.22, precision 0.67, hamming loss 0.1 and zero one loss 0.75 for the classification of the movie genre.