labelme: Image Polygonal Annotation with Python
===============================================
[](https://pypi.python.org/pypi/labelme)
[](https://travis-ci.org/wkentaro/labelme)
[](https://hub.docker.com/r/wkentaro/labelme)
Labelme is a graphical image annotation tool inspired by .
It is written in Python and uses Qt for its graphical interface.
Requirements
------------
- Ubuntu / macOS / Windows
- Python2 / Python3
- [PyQt4 / PyQt5](http://www.riverbankcomputing.co.uk/software/pyqt/intro)
Installation
------------
There are options:
- Platform agonistic installation: Anaconda, Docker
- Platform specific installation: Ubuntu, macOS
**Anaconda**
You need install [Anaconda](https://www.continuum.io/downloads), then run below:
```bash
# python2
conda create --name=labelme python=2.7
source activate labelme
conda install pyqt
pip install labelme
# python3
conda create --name=labelme python=3.6
source activate labelme
# conda install pyqt
pip install pyqt5 # pyqt5 can be installed via pip on python3
pip install labelme
```
**Docker**
You need install [docker](https://www.docker.com), then run below:
```bash
wget https://raw.githubusercontent.com/wkentaro/labelme/master/scripts/labelme_on_docker
chmod u+x labelme_on_docker
# Maybe you need http://sourabhbajaj.com/blog/2017/02/07/gui-applications-docker-mac/ on macOS
./labelme_on_docker static/apc2016_obj3.jpg -O static/apc2016_obj3.json
```
**Ubuntu**
```bash
# Ubuntu 14.04
sudo apt-get install python-qt4 pyqt4-dev-tools
sudo pip install labelme # python2 works
```
**macOS**
```bash
# macOS Sierra
brew install pyqt # maybe pyqt5
pip install labelme # both python2/3 should work
```
Usage
-----
### Annotation
Run `labelme --help` for detail.
```bash
labelme # Open GUI
labelme examples/tutorial/apc2016_obj3.jpg # Specify file
labelme examples/tutorial/apc2016_obj3.jpg -O examples/tutorial/apc2016_obj3.json # Close window after the save
labelme examples/tutorial/apc2016_obj3.jpg --nodata # Not include image data but relative image path in JSON file
labelme examples/tutorial/apc2016_obj3.jpg \
--labels highland_6539_self_stick_notes,mead_index_cards,kong_air_dog_squeakair_tennis_ball # Specify label list
labelme examples/tutorial/ # Open directory to annotate all images in it
```
The annotations are saved as a [JSON](http://www.json.org/) file. The
file includes the image itself.
### Visualization
To view the json file quickly, you can use utility script:
```bash
labelme_draw_json examples/tutorial/apc2016_obj3.json
```
### Convert to Dataset
To convert the json to set of image and label, you can run following:
```bash
labelme_json_to_dataset examples/tutorial/apc2016_obj3.json -o examples/tutorial/apc2016_obj3_json
```
It generates standard files from the JSON file.
- [img.png](examples/tutorial/apc2016_obj3_json/img.png): Image file.
- [label.png](examples/tutorial/apc2016_obj3_json/label.png): Int32 label file.
- [label_viz.png](examples/tutorial/apc2016_obj3_json/label_viz.png): Visualization of `label.png`.
- [label_names.txt](examples/tutorial/apc2016_obj3_json/label_names.txt): Label names for values in `label.png`.
Note that loading `label.png` is a bit difficult
(`scipy.misc.imread`, `skimage.io.imread` may not work correctly),
and please use `PIL.Image.open` to avoid unexpected behavior:
```python
# see examples/tutorial/load_label_png.py also.
>>> import numpy as np
>>> import PIL.Image
>>> label_png = 'examples/tutorial/apc2016_obj3_json/label.png'
>>> lbl = np.asarray(PIL.Image.open(label_png))
>>> print(lbl.dtype)
dtype('int32')
>>> np.unique(lbl)
array([0, 1, 2, 3], dtype=int32)
>>> lbl.shape
(907, 1210)
```
Screencast
----------
Acknowledgement
---------------
This repo is the fork of [mpitid/pylabelme](https://github.com/mpitid/pylabelme),
whose development has already stopped.