This script can transffer images/labels from nuimages dataset to the form that yolo requires.
- Download the Nuscenes dataset, where you can access from https://nuscenes.org/
- Download the nuscenes dev-kit, where you can access from https://github.com/nutonomy/nuscenes-devkit
- Download the script from this repo
- Create the directories in which you want to store the result, eg: |_ yolo-dataset |_ images |_ train |_ valid |_ labels |_ train |_ valid
- --nu-root,where you put the nuscenes dataset.
- --yl-root, where you put the yolo directories that created above.
- --data-version, which type of data it is, eg train/val/test/mini. If you're not familiar with the operation, please try mini at first.
- --data-type, the target directory name in point4 above, eg train/valid.
- --processor, number of threads to deal with the data, depends on your hardware.