Rashid Azim et al. (2026). An explainable deep learning framework for video violence detection using unsupervised keyframe selection and attention-based CNN. Scientific Reports. https://doi.org/10.1038/s41598-026-40977-7