Image annotation tool with added support for DICOMs
Clone or download
Pull request Compare This branch is 30 commits ahead, 2 commits behind tzutalin:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
build-tools
data
demo
icons
libs
scripts
tests
.gitignore
.travis.yml
CONTRIBUTING.rst
HISTORY.rst
LICENSE
MANIFEST.in
Makefile
README.md
RunLabelImg
__init__.py
environment.yml
issue_template.md
labelImg.py
resources.qrc
setup.cfg
setup.py

README.md

LabelImg

Forked from the LabelImg repo by tzutalin. Updated to work with DICOMs.

Installation

Mac OS X and Linux (Ubuntu 16.04)

  1. Make sure you have Anaconda or Miniconda installed.
  2. Clone this repo and cd into the top-level folder.
  3. Run the following
    conda env create -f environment.yml
    conda activate label-img
    make qt4py2
    python labelImg.py
    

Usage

Annotate

  1. Open Terminal, cd into the label-img directory
  2. Run conda activate label-img
  3. Run python labelImg.py
  4. Click 'Change default saved annotation folder' in Menu/File
  5. Click 'Open Dir'
  6. Click 'Create RectBox'
  7. Click and release left mouse to select a region to annotate the rect box
  8. You can use right mouse to drag the rect box to copy or move it

The annotation will be saved to the folder you specify.

Make sure PascalVOC (not YOLO) is selected in the toolbar on the left-hand side.

You can refer to the below hotkeys to speed up your workflow.

Create pre-defined classes

You can edit the file in data/predefined_classes.txt to define your own classes.

Keyboard Shortcuts

Keys Description
w Create a bounding box
d Next image
a Previous image
cmd + scroll Zoom in/out
cmd + s Save
cmd + d Copy the current label and bounding box
del Delete the selected bounding box
space Mark the current image as verified
cmd + '+' Zoom in
cmd + '-' Zoom out
↑ → ↓ ← Move selected bounding box

License

Free software: MIT license

Citation: Tzutalin. LabelImg. Git code (2015). https://github.com/tzutalin/labelImg