bmxbmx3 / annotation_transform_pyqt

transform annotation format for CV learning, like yolo, coco, etc.

annotation_transform_pyqt

Transform annotation formats for CV learning, like yolo, coco, etc.

Environment

python >= 3.7
Pillow >= 8.4.0
PyQt5 >= 5.15.6

How to use

Just run the command python main.py,then it will show the pyqt window:

main window

What it can do

1. Transform the common annotation formats for CV learning.
2. Rename, filter in/out the label name.

The core idea to implement this code

concept

We can use the compilation thought to make a middle format called "middle json", which can connect any two of the common used annotation formats, like yolo or coco.

The basic "middle json" format is like below:

{
  "image": {
    "path": "",
    "width": int,
    "height": int,
    "extra": {}
  },
  "label": [
    {
      "name": "",
      "xmin": int,
      "xmax": int,
      "ymin": int,
      "ymax": int,
      "extra": {}
    }
  ]
}

If you want to extend more formats, just follow the code in the "algorithm" directory to make your own. And I'm glad to welcome your PR for this project. :)

annotation_transform_pyqt's Issues

第一次使用需要先在algorithm目录下创建middle_json文件夹，要不然会报错