一言难尽的数据标注

阅读量：7068 次

发布时间：2019-06-28

本文共 5284 字，大约阅读时间需要 17 分钟。

LabelImg

LabelImg is a graphical image annotation tool.

It is written in Python and uses Qt for its graphical interface.

Annotations are saved as XML files in PASCAL VOC format, the format used by . Besides, it also supports YOLO format

Installation

Build from source

Linux/Ubuntu/Mac requires at least and has been tested with . However, and are strongly recommended.

Ubuntu Linux

Python 2 + Qt4

sudo apt-get install pyqt4-dev-toolssudo pip install lxmlmake qt4py2python labelImg.pypython labelImg.py [IMAGE_PATH] [PRE-DEFINED CLASS FILE]

Python 3 + Qt5 (Recommended)

sudo apt-get install pyqt5-dev-toolssudo pip3 install -r requirements/requirements-linux-python3.txtmake qt5py3python3 labelImg.pypython3 labelImg.py [IMAGE_PATH] [PRE-DEFINED CLASS FILE]

macOS

Python 2 + Qt4

brew install qt qt4brew install libxml2make qt4py2python labelImg.pypython labelImg.py [IMAGE_PATH] [PRE-DEFINED CLASS FILE]

Python 3 + Qt5 (Recommended)

brew install qt  # Install qt-5.x.x by Homebrewbrew install libxml2or using pippip3 install pyqt5 lxml # Install qt and lxml by pipmake qt5py3python3 labelImg.pypython3 labelImg.py [IMAGE_PATH] [PRE-DEFINED CLASS FILE]

Python 3 Virtualenv (Recommended)

Virtualenv can avoid a lot of the QT / Python version issues

brew install python3pip3 install pipenvpipenv --three # or pipenv install pyqt5 lxmlpipenv run pip install pyqt5 lxmlpipenv run make qt5py3python3 labelImg.py[Optional] rm -rf build dist; python setup.py py2app -A;mv "dist/labelImg.app" /Applications

Note: The Last command gives you a nice .app file with a new SVG Icon in your /Applications folder. You can consider using the script: build-tools/build-for-macos.sh

Windows

Install , and .

Open cmd and go to the directory

pyrcc4 -o line/resources.py resources.qrcFor pyqt5, pyrcc5 -o libs/resources.py resources qrcpython labelImg.pypython labelImg.py [IMAGE_PATH] [PRE-DEFINED CLASS FILE]

Windows + Anaconda

Download and install (Python 3+)

Open the Anaconda Prompt and go to the directory

conda install pyqt=5pyrcc5 -o libs/resources.py resources.qrcpython labelImg.pypython labelImg.py [IMAGE_PATH] [PRE-DEFINED CLASS FILE]

Get from PyPI but only python3.0 or above

pip3 install labelImglabelImglabelImg [IMAGE_PATH] [PRE-DEFINED CLASS FILE]

Use Docker

docker run -it \--user $(id -u) \-e DISPLAY=unix$DISPLAY \--workdir=$(pwd) \ --volume="/home/$USER:/home/$USER" \ --volume="/etc/group:/etc/group:ro" \ --volume="/etc/passwd:/etc/passwd:ro" \ --volume="/etc/shadow:/etc/shadow:ro" \ --volume="/etc/sudoers.d:/etc/sudoers.d:ro" \ -v /tmp/.X11-unix:/tmp/.X11-unix \ tzutalin/py2qt4 make qt4py2;./labelImg.py

You can pull the image which has all of the installed and required dependencies.

Usage

Steps (PascalVOC)

Build and launch using the instructions above.

Click 'Change default saved annotation folder' in Menu/File

Click 'Open Dir'

Click 'Create RectBox'

Click and release left mouse to select a region to annotate the rect box

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.

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

Steps (YOLO)

In data/predefined_classes.txt define the list of classes that will be used for your training.

Build and launch using the instructions above.

Right below "Save" button in the toolbar, click "PascalVOC" button to switch to YOLO format.

You may use Open/OpenDIR to process single or multiple images. When finished with a single image, click save.

A txt file of YOLO format will be saved in the same folder as your image with same name. A file named "classes.txt" is saved to that folder too. "classes.txt" defines the list of class names that your YOLO label refers to.

Note:

Your label list shall not change in the middle of processing a list of images. When you save an image, classes.txt will also get updated, while previous annotations will not be updated.

You shouldn't use "default class" function when saving to YOLO format, it will not be referred.

When saving as YOLO format, "difficult" flag is discarded.

Create pre-defined classes

You can edit the to load pre-defined classes

Hotkeys

Ctrl + u	Load all of the images from a directory
Ctrl + r	Change the default annotation target dir
Ctrl + s	Save
Ctrl + d	Copy the current label and rect box
Space	Flag the current image as verified
w	Create a rect box
d	Next image
a	Previous image
del	Delete the selected rect box
Ctrl++	Zoom in
Ctrl--	Zoom out
↑→↓←	Keyboard arrows to move selected rect box

Verify Image:

When pressing space, the user can flag the image as verified, a green background will appear. This is used when creating a dataset automatically, the user can then through all the pictures and flag them instead of annotate them.

Difficult:

The difficult field is set to 1 indicates that the object has been annotated as "difficult", for example, an object which is clearly visible but difficult to recognize without substantial use of context. According to your deep neural network implementation, you can include or exclude difficult objects during training.