Development of a Violence Scene Prediction System in Films Based on Visual Poster Analysis Using The Machine Learning Method Convolutional Neural Network(CNN)

Immanuel V William (Universitas Teknologi Yogyakarta)
Ahmad Tri Hidayat (Universitas Teknologi Yogyakarta)

Journal of Scientific Research, Education, and Technology
Vol. 5 No. 2 (2026): Vol. 5 No. 2 2026

Films as an entertainment often contain various visual elements, one of which is violent scenes. In some cases, the presence of violence in films can have a negative impact on viewers, especially children. Therefore, a system is needed that can predict the potential existence of violent scenes in a film, particularly based on poster. This research aims to develop a system that uses machine learning methods to analyze movie posters and estimate the likelihood of the film containing violent scenes. The designed system will be trained using a dataset of movie posters labeled according to the level of violence present in the films. The posters will be analyzed using digital image processing techniques, and the results will serve as input for the machine learning model used in the system. The development process involves using artificial neural network (ANN) techniques, which have been proven effective in detecting relevant visual patterns. The research results show that the proposed system can achieve a high level of accuracy in predicting films that contain violent scenes based on their visual posters. Thus, this system can be useful for the film industry, parents, and regulatory agencies to identify inappropriate films for certain age before their release.