Vision Dataset, is a collection of detailed annotations, that has been generated with the purpose of aiding semantic segmentation and related scene understanding problems in the context of driving scenarios. The datasets comes with precise pixel-level semantic annotations.