---
license: mit
task_categories: [image-segmentation]
---

# Cityscapes Dataset
The Cityscapes dataset is a large-scale dataset designed for urban scene understanding, with a focus on semantic segmentation, instance segmentation, and panoptic segmentation. It contains high-quality pixel-level annotations for images captured in 50 different cities, making it a benchmark for evaluating computer vision models in real-world urban environments.