CINEVISIONAI: A MULTI-MODAL DEEP LEARNING FRAMEWORK FOR MOVIE SUCCESS AND RATING PREDICTION USING DATA MINING

Abstract

The film industry represents a high-risk domain where massive financial investments are made without guaranteed returns. Predicting the success rate of a movie prior to its release has become a significant research area in machine learning and artificial intelligence. This paper presents a Convolutional Neural Network (CNN)-based model to predict the success rate of movies using features such as posters, trailers, cast information, genre, budget, and audience sentiment. Traditional prediction models rely heavily on structured numerical data and statistical approaches, which often fail to capture the visual and emotional appeal of a movie. By combining image-based analysis with metadata and textual sentiment analysis using CNN along with LSTM, RNN, Random Forest, and other classifiers, the proposed system achieves superior prediction accuracy of 98.32% compared to existing methods. The system classifies movies into categories such as Hit, Average, or Flop, assisting producers, distributors, and investors in making informed decisions and minimizing financial risks.