🛠️ Label checking tool 🛠️

Description ℹ️

Label validation tool written in python.
The script draws bounding boxes on corresponding images and enables users to navigate through the dataset easily using the keyboard.
Images not related to any label are automaticaly send to incorrect data folderd, as are labels.
Supported image formats: .png, .jpg, .jpeg, .bmp.
Labels schould be stored in .txt files in YOLO format.

Requirements ⚙️

You need to install opencv library for python:

pip install opencv-python

Usage 👈

You can drop check_labels.py in folder containing your data (both images and labels) and run it via console:

python check_labels.py 

Youn can also send args containing paths to images, labels, incorrect data folder and image format.

python check_labels.py -ip ./images/ -lp ./labels/ -icp ./IncorrectData/ -if .png

Keybinding ⌨️

• Next image d
• Previous image a
• Send to incorrect data folder Space
• Quit q

Default parameters 🧰

• Path to labels: './'
• Path to images: './'
• Path to incorrect data folder: './IncorrectData/'
• Image format: '.png'
• Size of displayed image: 1280x736
• Labels format: '.txt'

License 📄

MIT License