This is a small tool that I programmed to efficiently label data for a classification task.
This was made for Structured Light data, but can be easily repurposed for arbitrary images.
Don't expect anything too much here, lol.
To get started with the labelling tool, follow these steps:
-
Clone the repository:
git clone https://github.com/Henningson/FastLabeling.git cd FastLabeling
-
Install dependencies:
pip install numpy pyqt5 opencv-python
-
Run the tool:
python main.py
1
- Label as Laserpoint2
- Label as Specular Highlight3
- Label as Other
Labels are then highlighted in the following colors:
- 🟩 Laserpoint
- 🟥 Specular Highlight
- 🟦 Other
- ⬛ Not yet labeled
-
Open Image Folder:
- Use the dropdown menu to navigate to and select the folder containing your images.
-
Label Images:
- Use the key-bindings (
1
,2
,3
) to label images as Laser Dot, Specular Highlight, or Other. The GUI will automatically advance to the next image after each label is applied.
- Use the key-bindings (
-
Save Progress:
- Use the save dropdown to save your current labelling progress to a specified path.
-
Generate Dataset:
- Use the generate dataset option to split your labelled data into train, validation, and test sets (80/10/10 split). Note that unlabeled data is ignored in this step. The split can be changed in the code itself.
This project is licensed under the MIT License. See the LICENSE file for more details.