https://github.com/Setogit/labelimg-data-converter

Converts labelimg meta data to yolo format. See https://github.com/tzutalin/labelImg
https://github.com/Setogit/labelimg-data-converter

Last synced: about 2 months ago
JSON representation

Converts labelimg meta data to yolo format. See https://github.com/tzutalin/labelImg

• Host: GitHub
• URL: https://github.com/Setogit/labelimg-data-converter
• Owner: Setogit
• License: mit
• Created: 2018-02-13T19:27:48.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-04-05T20:44:19.000Z (over 6 years ago)
• Last Synced: 2024-06-10T12:44:41.074Z (3 months ago)
• Language: Python
• Size: 14.6 KB
• Stars: 1
• Watchers: 1
• Forks: 1
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# labelimg-data-converter

Converts labelimg meta data to yolo format.  See https://github.com/tzutalin/labelImg

```

$ python labelimg_data_converter.py -h

usage: python labelimg_data_converter.py  such as "cat,dog,horse,pig"

Converts labelimg meta data to yolo meta data. The original image files

(*.jpg) and lalbelimg meta data XML files (*.xml) must exist in sub-

directories named "movie" where  is an positive integer number. In

the sub-directories, the jpg and xml files must exist with file names as

"frame.jpg" and "frame.xml where  is either 3- or 4-digit

number. For example, /movie1/frame001.xml or

/movie123456/frame1234.xml

positional arguments:

  classes               [REQUIRED] a comma delimited list of

                        class_name:class_id pairs. space characters are

                        preserved. e.g., "dog:1,cat:0,horse:3,peter rabbit:5"

                        is translated to {"cat": 0, "dog": 1, "horse": 3,

                        "peter rabbit": 5}

optional arguments:

  -h, --help            show this help message and exit

  -s SRC, --source SRC  directory where the original labelimg meta data XML

                        and image JPG files resides. default: "source"

  -d DEST, --destination DEST

                        directory where the sanitized and converted yolo meta

                        data TXT and image JPG as well as the original

                        labelimg meta data XML will be stored. default:

                        "destination"

  -b SUBDIR, --subdir SUBDIR

                        sub-directory where the generated files are stored

                        beneath the destination directory. default: "data"

  -e HEADER, --header HEADER

                        string used as header of the generated yolo meta data

                        TXT file names. default: "sample"

  -p {(0.0 .. 1.0) exclusive}, --percentage_test {(0.0 .. 1.0) exclusive}

                        float number between 0.0 and 1.0 exclusive to specify

                        the amount of labeled images to be used for validation

                        in training. default: 0.1, which means 10 percent.

```