Skip to content

This repo contains usable code convert cityscapes to coco format (Detectron and maskrcnn-benchmark were all broken)

Notifications You must be signed in to change notification settings

wxyz-lang/cityscapestococo

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 

Repository files navigation

cityscapestococo

This repo contains all code which convert cityscapes to coco. The scripts actually provided inside Detectron and maskrcnn-benchmark were all down. To help others quickly setup their training pipeline on cityscapes for instance segmentation, this repo is really helpful.

The converted result cityscapes to coco visualized using vis_coco.py inside this repo:

As you can see, all instance classes were token out from cityscapes, which can be conclude as:

[
        'person',
        'rider',
        'car',
        'truck',
        'bus',
        'train',
        'motorcycle',
        'bicycle',
    ]

Usage

python3 convert_cityscapes_to_coco.py

For visual:

python3 vis_coco.py path/to/annn.json path/to/images/

note: You have to move all cityscapes images to a single folder without any subfolders just follow the coco structures.

Copyright

All rights belongs to Fagang Jin, codes released under Apache License

About

This repo contains usable code convert cityscapes to coco format (Detectron and maskrcnn-benchmark were all broken)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 99.8%
  • C 0.2%