A dataset can be used by accessing DatasetCatalog
for its data, or MetadataCatalog for its metadata (class names, etc).
This document explains how to setup the builtin datasets so they can be used by the above APIs.
Use Custom Datasets gives a deeper dive on how to use DatasetCatalog and MetadataCatalog,
and how to add new datasets to them.
CLOUDS has builtin support for a few datasets, the datasets are assumed to exist in datasets/ directory.
Under this directory, detectron2 will look for datasets in the structure described below.
Cityscapes: Please, download leftImg8bit_trainvaltest.zip and
gt_trainvaltest.zip from here
and extract them to datasets/cityscapes.
GTA: Please, download all image and label packages from
here and extract
them to datasets/gta5.
Synthia: Please, download SYNTHIA-RAND-CITYSCAPES from
here and extract it to datasets/synthia.
ACDC: Please, download rgb_anon_trainvaltest.zip and
gt_trainval.zip from here and
extract them to datasets/acdc.
BDD100K: Please, download the 10K Images and Segmentation from
here and extract it
to datasets/bdd100k.
Mapillary: Please, download the mapillary-vistas-dataset_public_v1.2.zip
from here and extract it
to datasets/mapillary_vistas.
The final folder structure should look like this:
HRDA
├── ...
├── datasets
│ ├── acdc_list
│ ├── acdc
│ ├── bdd_list
│ ├── bdd100k
│ ├── cityscapes_list
│ ├── cityscapes
│ ├── gta_list
│ ├── gta
│ ├── mapillary_list
│ ├── mapillary_vistas
│ ├── synthia_list
│ ├── synthia
├── ...
Data Preprocessing: Finally, please run the following scripts to convert the label IDs to the train IDs:
python tools/convert_datasets/gta.py data/gta --nproc 8
python tools/convert_datasets/cityscapes.py data/cityscapes --nproc 8
python tools/convert_datasets/synthia.py data/synthia/ --nproc 8
python tools/convert_datasets/mapillary.py data/mapillary/ --nproc 8