ultralytics-main_2

This dataset combines a curated subset of the COCO dataset (COCO128) and a custom traffic scene dataset containing 2,704 annotated images. COCO128 provides a standardized baseline for evaluating object detection models, while the traffic dataset focuses on real-world urban road scenarios, including vehicles, pedestrians, traffic signs, and road markings. The combined dataset supports rapid prototyping and robust evaluation of object detection models across general and domain-specific tasks. All images are annotated in YOLO format, making it directly compatible with Ultralytics YOLO implementations and other popular detection frameworks.