Genre Classification of Movie Trailers Using Audio and Visual Features: A Comparative Study of Machine Learning Algorithms

Movie trailers are a crucial marketing tool for the film industry and are often used to generate audience interest and anticipation. Automatic genre classification of movie trailers can assist filmmakers in targeting their intended audience and help viewers in deciding which films to watch. This research paper aims to investigate the effectiveness of various machine learning algorithms for the classification of movie genres based on audio and visual features extracted from movie trailers. We compare the performance of several classifiers, including Support Vector Machines (SVM), Random Forest (RF), Naive Bayes (NB), and K- Nearest Neighbors (KNN) on a dataset of movie trailers belonging to five different genres—action, comedy, drama, horror, and thriller. We extract both audio and visual features from the trailers, including spectrogram features, pitch, loudness, brightness, contrast, and color histograms. We then use these features to train and evaluate the different classifiers. Moreover, we observed that combining both audio and visual features improves the overall accuracy of genre classification. Our study contributes to the field of movie genre classification by providing a comparative analysis of different machine learning algorithms for the classification of movie trailers based on both audio and visual features. The findings of this research can be applied in various domains, such as movie recommendation systems, marketing strategies, and content analysis.