Lightweight Framework for Railway Obstacle Detection using YOLO Network

The development of a real-time railway obstacle detection framework based on an optimized YOLO network to ensure increased safety and operational efficiency by accurate identification of small objects such as debris, animals, and other possible hazards on railway tracks. In this research, there are two groups. Group 1 included a sample size of 50 images and videos using the existing Mask R-CNN algorithm, with a detection accuracy of 76.87%. Group 2 included a sample size of 50 and was developed using the optimized YOLO framework and achieved 95% accuracy. Both groups used datasets obtained from Kaggle and Roboflow. Some key technologies are based on Python modeling, HTML, CSS, MySQL and WampServer, Windows and a level of Significance less than 0.050 The proposed YOLO-based framework completed the tasks with a precision of 86.47% while surpassing the Mask R-CNN method at 76.87% as calculated with its real-time processing capacity of 60 FPS (0.0167 seconds per frame), easing the computational complexity by 42% when compared to Mask R-CNN .The optimized YOLO framework is very efficient for real-time railway obstacle detection, offering better accuracy and speed as compared to the traditional methods.