BBox-Label-Tool

A multi-class bounding box labeling tool

Improvements in this fork

Conversion script to convert bounding box labels to YOLO format
Bounding boxes are now labeled
Dropdown menu for loading directories
Image resizing script
No need to click "Confirm Class" to switch classes

Installation

Requirements

Python 3
Python3 Tkinter

Recommended: setup virtualenv

Install requirements

$ sudo apt-get install python3 python3-tk
$ pip3 install -r requirements.txt

Usage

Preparing the images

Create folders in Images/ and place your images in them, e.g. Images/Dogs/, Images/Cats/, etc. There is no strict requirement about what images need to be in which folders; this is just an organizational feature.

Running the labeling tool

$ python3 main.py

To change image directory, use the dropdown menu next to Image dir:
To change object class, use the dropdown menu under Class:
To draw a bounding box, use your mouse to align the crosshairs and click at two corner points
To cancel drawing a bounding box, press esc
To delete a bounding box, select the box in the list and click the Delete button
To change images, either use the left/right arrows on your keyboard, a or d, or click the buttons
Annotations are saved in the following format: <x1>, <y1>, <x2>, <y2>, <w>, <h>, <class>, where x1 y1 x2 and y2 are the top left and bottom right coordinates of the bounding box, w and h are the width and height of the images, and class is the category of the bounding box.

Figuring out the rest is left as an exercise to the reader.

Help! My image is too big to fit on the screen.

Run resize_images.py to resize all images to a specified max height. Default is 800px.

Converting to YOLO Format

$ python3 convert.py

Labels will be converted to YOLO format. Formatted labels will be saved in the specified folder, default is YOLO_Formatted/. Only images in the Images folder will have their labels converted.

andrewhu/Multiclass-BBox-Label-Tool