Labelme2YOLOv8

Forked from[ greatv/labelme2yolo](https://github.com/greatv/labelme2yolo)

Labelme2YOLOv8 is a powerful tool for converting LabelMe's JSON format to [YOLOv5](https://github.com/ultralytics/ultralytics) dataset format. This tool can also be used for YOLOv5/YOLOv8 segmentation datasets, if you have already made your segmentation dataset with LabelMe, it is easy to use this tool to help convert to YOLO format dataset.

New Features

• export data as yolo polygon annotation (for YOLOv8 segmentation)
• Existing Structure (YOLOv5 v7.0)

• YOLODataset
  • images
    • test
    • train
    • val
  • labels
    • test
    • train
    • val

• Updated Structure (YOLOv8)

• YOLOv8Dataset
  • test
    • images
    • labels
  • train

    • images

    • labels

  • val
    • images
    • labels

Installation

pip install labelme2yolov8

Arguments

--json_dir LabelMe JSON files folder path.

--val_size (Optional) Validation dataset size, for example 0.2 means 20% for validation.

--test_size (Optional) Test dataset size, for example 0.2 means 20% for Test.

--json_name (Optional) Convert single LabelMe JSON file.

--output_format (Optional) The output format of label.

--label_list (Optional) The pre-assigned category labels.

How to Use

1. Converting JSON files and splitting training, validation, and test datasets with --val_size and --test_size

You may need to place all LabelMe JSON files under labelme_json_dir and then run the following command:

labelme2yolo --json_dir /path/to/labelme_json_dir/ --val_size 0.15 --test_size 0.15

This tool will generate dataset labels and images with YOLO format in different folders, such as

/path/to/labelme_json_dir/YOLOv8Dataset/train/labels/
/path/to/labelme_json_dir/YOLOv8Dataset/test/labels/
/path/to/labelme_json_dir/YOLOv8Dataset/val/labels/
/path/to/labelme_json_dir/YOLOv8Dataset/train/images/
/path/to/labelme_json_dir/YOLOv8Dataset/test/images/
/path/to/labelme_json_dir/YOLOv8Dataset/val/images/

/path/to/labelme_json_dir/YOLOv8Dataset/dataset.yaml

2. Converting JSON files and splitting training and validation datasets by folders

If you have split the LabelMe training dataset and validation dataset on your own, please put these folders under labelme_json_dir as shown below:

/path/to/labelme_json_dir/train/
/path/to/labelme_json_dir/val/

This tool will read the training and validation datasets by folder. You may run the following command to do this:

labelme2yolo --json_dir /path/to/labelme_json_dir/

This tool will generate dataset labels and images with YOLO format in different folders, such as

/path/to/labelme_json_dir/YOLOv8Dataset/train/labels/
/path/to/labelme_json_dir/YOLOv8Dataset/val/labels/
/path/to/labelme_json_dir/YOLOv8Dataset/train/images/
/path/to/labelme_json_dir/YOLOv8Dataset/val/images/

/path/to/labelme_json_dir/YOLOv8Dataset/dataset.yaml

How to build package/wheel

1. install hatch
2. Run the following command:

hatch build

License

labelme2yolov8 is distributed under the terms of the MIT license.