This dataset contains 1028 images each 640x380 pixels.
The dataset is split into 249 test and 779 training examples.
Every image comes with MS COCO format annotations.
The dataset was collected in Carla Simulator, driving around in autopilot mode in various environments
(Town01, Town02, Town03, Town04, Town05) and saving every i-th frame.
The labels where then automatically generated using the semantic segmentation information.
