cityscapes

Cityscapes dataset

The Cityscapes dataset contains a diverse set of stereo video sequences recorded in street scenes from 50 different cities, with high quality pixel-level annotations of 5 000 frames in addition to a larger set of 20 000 weakly annotated frames. The dataset is thus an order of magnitude larger than similar previous attempts. Details on annotated classes and examples of our annotations are available at this webpage.

For more details, see: Cityscapes Dataset and Cityscapes Dataset at GitHub.

Prepare dataset

1. Download and unzip dataset files: Use csDownload or download the files mentioned below manually from: Cityscapes Dataset Downloads

   CITYSCAPES_DOWNLOAD_DIR="/path/where/to/store/cityscapes_downloads"
   
   # using cityscapesScripts
   # use "csDownload -l" to list available packages
   
   # labels (semantic, instance)
   csDownload gtFine_trainvaltest.zip -d $CITYSCAPES_DOWNLOAD_DIR    # -> 241MB
   # rgb images
   csDownload leftImg8bit_trainvaltest.zip -d $CITYSCAPES_DOWNLOAD_DIR     # -> 11GB
   # disparity images (only upon request)
   csDownload disparity_trainvaltest.zip -d $CITYSCAPES_DOWNLOAD_DIR     # -> 3.5GB
   # intrinsic and extrinsic camera parameter to calculate depth
   csDownload camera_trainvaltest.zip -d $CITYSCAPES_DOWNLOAD_DIR    # -> 2MB
   
   # unzip files
   find $CITYSCAPES_DOWNLOAD_DIR -name '*.zip' -exec unzip -o {} -d $CITYSCAPES_DOWNLOAD_DIR \;

2. Convert dataset:

   # general usage
   nicr_sa_prepare_dataset cityscapes \
       /path/where/to/datasets/cityscapes \
       $CITYSCAPES_DOWNLOAD_DIR