A Review of Research Trends in Movie Automation

The increasing reliance on data-driven tools in the film industry has led to a surge in research on automating key cinematic tasks such as content certification, genre classification, rating prediction, and commercial success forecasting. However, a consolidated understanding of how machine learning (ML) strategies, datasets, and training methodologies evolve across these tasks remains lacking. This survey bridges that gap by systematically reviewing 38 peer-reviewed IEEE papers published recently, offering a unique cross-sectional analysis across four subdomains of movie automation. IEEE papers were selected due to their rigorous peer review standards.Unlike previous surveys focused on individual applications, our study uniquely compares ML pipelines across five dimensions: dataset availability, data modality, algorithm type, training, and feature extraction strategies. We highlight temporal trends, identify underutilized practices (e.g., fine-tuning and combined use of traditional and deep algorithms) and uncover gaps such as poor reproducibility in custom-dataset studies. Through visual summaries, heatmaps, and Sankey diagrams, we provide practical insights into film-related ML research’s current state and future directions. Our findings offer a benchmarking foundation for researchers and practitioners, equipping them with the knowledge to develop scalable, reproducible, and intelligent film analytics systems.