yanwanquan / autolabelimg

auto-labelimg based on yolov5, with many other useful tools

AutoLabelImg：MultiFunction AutoAnnotate Tools

English | 中文

Introduction：

Based on labelImg, we add many useful annotate tools, in Annoatate-tools and Video-tools menu, including：

• TOOL LIST：
• Auto Annotate：anto annotate images using yolov5 detector
• Tracking Annotate：using tracking method in opencv, annotate video data
• Magnifing Lens：helpful when annotating small objects, optional function
• Data Agument：data agument
• Search System：search details info based on your input
• other tools：label selecting/rename/counting, fix annotation, video merge/extract, welcome to try

Demo:

seen in Vtuber:

Auto Annotate

Tracking Annotate

Magnifing Lens

Data Augment

Search System

Update log：

2022.01.14：remove Retinanet( matain yolov5 only), and add label selecting when autolabeling

2022.01.11：imporve magnifing lens, more fluent and can be shut

2020.12.28：add video tracking annotate

2020.12.10：autolabelimg，version 1.0

Installation：

1. clone this repo：

   git clone https://github.com/wufan-tb/AutoLabelImg
   cd AutoLabelImg

2. install requirments：

   conda create -n {your_env_name} python=3.7.6
   conda activate {your_env_name}
   pip install -r requirements.txt

3. compile source code：

   Ubuntu User:

   sudo apt-get install pyqt5-dev-tools
   make qt5py3
   

   Windows User:

   pyrcc5 -o libs/resources.py resources.qrc
   

4. prepare yolov5 weights file and move them to here: [official model zoo:Yolov5]

   mv {your_model_weight.pt} pytorch_yolov5/weights/

5. open labelimg software

   python labelImg.py
   

Set shortcut to open software[optional]

Windows User:

create a file:labelImg.bat, open it and type these text(D disk as an example)：

D:
cd D:{path to your labelImg folder}
start python labelImg.py
exit

double click labelImg.bat to open the software.

Ubuntu User:

open environment setting file:

vim ~/.bashrc

add this command：

alias labelimg='cd {path to your labelImg folder} && python labelImg.py

source it：

source ~/.bashrc

typing 'labeling' in terminal to open the software.

Citation

{   AutoLabelImg,
    author = {Wu Fan},
    year = {2020},
    url = {\url{https://https://github.com/wufan-tb/AutoLabelImg}}
}