CNN-Based Video Genre Classification System Using Frame Extraction, Visual Features, and Deep Learning Techniques

This work presents a CNN-based approach for automated video genre classification to address the growing need for scalable multimedia organization. Key frames are extracted from videos, preprocessed, and used to train a CNN model capable of learning visual patterns relevant to genre prediction. Optimization techniques such as data augmentation, batch normalization, and dropout are applied to improve generalization and reduce overfitting. The model is evaluated using standard performance metrics including accuracy, precision, recall, and F1-score, demonstrating clear improvements over traditional machine learning and handcrafted feature-based methods. Experimental results show that the proposed approach achieves reliable and consistent classification performance across multiple video genres. The system demonstrates strong potential for integration into real-world applications, including personalized recommendation systems, automated metadata generation, content indexing, and large-scale digital media management.