Overhead Contact System

This dataset contains annotated images of railway overhead contact systems (OCS) for object detection research. The dataset is designed for training and validating deep learning models, particularly YOLO-based detectors, to identify and localize key components in railway pantograph-catenary systems. Dataset Structure: images/train: 11,009 training images images/val: 3,145 validation images labels/: YOLO-format annotation files (.txt) corresponding to each image, containing normalized bounding box coordinates and class labels Total: 14,154 images with pixel-level annotations for component detection tasks including insulators, brackets, and other OCS hardware. The dataset was collected from real-world railway inspection scenarios and annotated for computer vision research in railway infrastructure maintenance and automated defect detection.