Convolutional Neural Network based Railway Track Hazard Detection System

Abstract:- Ensuring railway safety is critical for preventing accidents and saving lives. This paper proposes an advanced deep learning-based system for real-time hazard detection on railway tracks. Using pre-trained Convolutional Neural Networks (CNN), including ResNet152V2, EfficientNetB7, and InceptionV3, the system identifies obstacles and recognizes railway signals. Extensive experiments on custom datasets demonstrate robust performance under various conditions, achieving a detection accuracy of over 90%. The proposed system is scalable for real-world applications with promising results in deployment scenarios.