DATA.md

Dataset preparation

We recommend creating a symlink for your dataset root at $GSS/data. If your directory structure varies, please adjust the relevant paths in the configuration files for proper alignment.

Data folder structure

mmsegmentation
├── mmseg
├── tools
├── configs
├── data
│   ├── cityscapes
│   │   ├── leftImg8bit
│   │   │   ├── train
│   │   │   ├── val
│   │   ├── gtFine
│   │   │   ├── train
│   │   │   ├── val
│   ├── ade
│   │   ├── ADEChallengeData2016
│   │   │   ├── annotations
│   │   │   │   ├── training
│   │   │   │   ├── validation
│   │   │   ├── images
│   │   │   │   ├── training
│   │   │   │   ├── validation
│   ├── mseg_dataset
│   │   ├── ADE20K
│   │   ├── Cityscapes
│   │   ├── KITTI
│   │   ├── PASCAL_VOC_2012
│   │   ├── WildDash
│   │   ├── BDD
│   │   ├── COCOPanoptic
│   │   ├── MapillaryVistasPublic
│   │   ├── ScanNet
│   │   ├── Camvid
│   │   ├── IDD
│   │   ├── PASCAL_Context
│   │   ├── SUNRGBD           

Cityscapes

You can access the data here once you've registered.

**labelTrainIds.png is utilized for Cityscapes training. MMSeg have provided a script, built upon cityscapesscripts, to generate the **labelTrainIds.png files.

# --nproc means 8 process for conversion, which could be omitted as well.
python tools/dataset_converters/cityscapes.py data/cityscapes --nproc 8

ADE20K

You can download the training and validation sets for ADE20K from this link. Additionally, the test set is available for download here.

MSeg

Please follow MSeg download instruction to download MSeg dataset