YOLOv8-Pyside6-GUI V2.4-alpha a GUI for Ultralytics

---

English   |   繁體中文

How to Run

1. Create virtual environment

Create a virtual environment equipped with python version 3.10, then activate environment.

conda create --name yolov8-gui python=3.10
conda activate yolov8-gui

2. Install dependency packages

pip install pyside6
pip install chardet
pip install pytube
pip install ultralytics==8.3.90
pip install torch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2 --index-url https://download.pytorch.org/whl/cu118

Change other pytorch version in

3. Run Program

• Traditional Chinese

python main.py

• English

There is an error when performing video prediction with the English version, and the cause is currently unknown.

python main_en.py

Packaging

1. Create virtual environment

It is recommended to use native Python to create a virtual environment and install Python 3.10.

python -m venv venv
(your YOLOv8-GUI-PySide6-main PATH)\venv\Scripts\activate.bat

2. Install dependency packages

pip install pyside6
pip install chardet
pip install pytube
pip install ultralytics==8.3.90
pip install torch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2 --index-url https://download.pytorch.org/whl/cu118
pip install lapx
pip install auto-py-to-exe

Change other pytorch version in

3. Launch the auto-py-to-exe UI

auto-py-to-exe

4. Add Script Location and Additional Files

Script Location

(your YOLOv8-GUI-PySide6-main PATH)\main.py

Add additional directories by clicking "Add Folder".

(your YOLOv8-GUI-PySide6-main PATH)\venv\Lib\site-packages\ultralytics

Click "Convert".

5. Copy Files

Copy the config, img, models, ui, and utils directories to (your YOLOv8-GUI-PySide6-main PATH)\output\main.

6. Launch main.exe

Run main.exe to start the application.

Notice

• ultralytics follows the GPL-3.0, if you need commercial use, you need to obtain its license.
• If you expect to use your own model, you need to use ultralytics to train the yolov5/6(only det)/8/9(det&seg)/10(only det)/11/12 model first, and then put the trained .pt file into the models/* folder.
• There are still some bugs in the software, and I will continue to optimize and add some more interesting functions as my time allows.
• If you check the save results, they will be saved in the ./run path
• The UI design file is home.ui, if you modify it, you need to use the pyside6-uic home.ui > ui/home.py command to regenerate the .py file
• The resource file is resources.qrc, if you modify the default icon, you need to use the pyside6-rcc resources.qrc > ui/resources_rc.py command to regenerate the .py file
• In the Oriented bounding box (obb) mode in Detect mode, if you want to use your own trained obb model, you need to add obb to the filename, such as yolov8n-obb.pt. If "obb" is not added, it will only enter the general detection mode.

Existing Features

1. Mode Selection

• Image Classification
• Object Detection
• Object Detection(OBB)
• Pose Estimation
• Instance Segmentation
• Object Tracking

2. Data Input Methods

• Single File Detection
• Folder(batch) detection function
• Support Dragging File Input
• Camera Input Support
• Support for 'chose_rtsp' and 'load_rtsp' Functions

To Do

• Monitor system hardware usage
• Graph showing changes in target quantity

Using Frameworks

References

• PyQt5-YOLOv5
• ultralytics
• PySide6-YOLOv8
• YOLOSHOW