The traditional track surface defect detection adopts the method of manual inspection which is subjective inefficient and time-consuming In recent years the detection method based on computer vision a

These studies have shown the potential of computer vision and image processing technology in improving the accuracy and efficiency of track surface defect detection in the railway industry. By combining deep learning algorithms with traditional detection methods, such as support vector machines and morphological operations, researchers have been able to achieve more effective and reliable defect detection.

The use of improved lightweight detection networks and cascade detection methods allows for faster and more efficient detection, making these methods suitable for on-line detection in real-time scenarios. Additionally, the ability to handle unbalanced samples and achieve defect detection for both straight and curved rails further enhances the versatility and practicality of these approaches.

Furthermore, the integration of multiple detection tasks, such as rail fasteners and rail surface defects, into a single network demonstrates the potential for comprehensive defect detection systems that can address different types of defects simultaneously.

Overall, these advancements in computer vision and image processing technology have significantly improved the efficiency and accuracy of track surface defect detection in the railway industry, paving the way for more automated and reliable inspection systems