Article
EFFICIENT RAILWAY FOREIGN OBJECT DETECTION USING YOLOV8
Maintaining the security and dependability of rail transportation systems depends on the prompt and precise detection of foreign items on railway lines. By combining a two-stage architecture based on YOLOv8 and Overhaul Knowledge Distillation (OKD), this study offers an improved foreign object intrusion detection framework that solves the drawbacks of current approaches, specifically low efficiency and suboptimal accuracy. To lessen the computational load on detection models, a lightweight image classification model quickly filters railway photos to find those that might include foreign items. The YOLOv8 object detector precisely locates and identifies the foreign objects in images that have been marked as suspicious. YOLOv8 provides notable improvements over its predecessors in terms of inference speed and detection accuracy. The Overhaul Knowledge Distillation technique is used to further improve the classification stage's performance, enabling the lightweight classifier to learn from a more intricate teacher network and attain competitive accuracy with increased efficiency. The suggested method establishes a new state-of-the-art in railway foreign item detection by outperforming current solutions in both speed and robustness, according to experimental evaluations.
Full Text Attachment
