LabelCraft

A simple yet powerful Tkinter-based GUI tool to create, edit, and export bounding box annotations in YOLO format for image datasets.
Ideal for training YOLO-based object detection models.

---

Application GUI

How to Start With this application

Step 1:

• option 1 : Download the repo
• Option 2 : Clone git repo

      git clone https://github.com/abyshergill/Label_Craft.git

  • Make Sure python Python ≥ 3.8

Step 2:

• Install the necessary library

      pip install requirments.txt

• Recommneded : Create Virtual environment and install all the necessary libaries inside the virutal environment.

Step 3:

• Run the application

      python main.py

How to Use

• Load image folder and automatically detect all supported formats (.jpg, .jpg, .bmp, .tiff, etc.)

  

• Add, remove, or rename object classes dynamically.

  • Active Class will be the class for bounding box drawing.
  • Keyboard Shortcuts
    • Key Action
    • Left Arrow Previous Image
    • Right Arrow Next Image
    • 0-9 Switch active class by index

  

  

• Draw bounding boxes directly on images using your mouse. You can annotate single as well as multi class.

  

• Navigate between images easily with “Next” and “Previous” buttons or arrow keys.

• Auto-save and export annotations in standard YOLO format: class_id x_center y_center width height

  • All annotation will be same with in folder as image_name.txt
  • Built-in support for existing annotations – loads .txt files if present.

• Displays annotations with distinct colors per class.

• Exports classes.txt file automatically for YOLO training.

Note : All the annotation file will be stored in annotation image folder.

---

Ultralytic YOLO format

## If you have single annotation in one image
<class_id> <x_center> <y_center> <width> <height>

## If you have multiple annotation in one image
<class_id> <x_center> <y_center> <width> <height>
<class_id> <x_center> <y_center> <width> <height>
<class_id> <x_center> <y_center> <width> <height>
<class_id> <x_center> <y_center> <width> <height>

Author Information

• Name : Kuldeep Singh
• Email : shergillkuldeep@outlook.com
• Linkedin :
• GitHub : https://github.com/abyshergill

License

This project is released under the GPL License

Future Plan

In future i will add more annotation format and different folder selection option during export.

⭐ Acknowledgments

Inspired by popular labeling tools like LabelImg and Roboflow Annotator, but designed to be lightweight, offline, and open-source.