Python-Yolo-Wrapper-for-Unity-interactions

Introduction

This source code has been developped to allow python and these libraries communicate with Unity Engine. This use case is using Ultralytics's YoloV8 and is able to send position information to unity in order to create interactions and animations with it.

Prerequisites

• Unity 2022.3.13f1 or later
• Python 3.8 or later
• Make sure you have a webcam or a video file to test the project
• This installation guide has been tested on Linux and MacOS, but it should work on Windows too with some modifications

Repository layout

.
├── configs/               # Tracker configuration presets
├── docs/assets/           # Images and plots used in the documentation
├── scripts/               # Helper scripts (e.g. Windows installer)
├── src/pywui/             # Python package source code
├── tests/                 # Automated tests
├── Makefile               # Common development tasks
├── pyproject.toml         # Build system and packaging metadata
└── UnityProject/          # Unity sample project

Installation

⚠️ if you have a windows PC with RTX GPU and Intel CPU check the follow installation guide ⚠️

• Lien de la documentation

Clone this public repo with wherever you want

git clone https://github.com/mathis-lambert/python-yolo-wrapper-for-unity-interactions yolo_with_unity
cd yolo_with_unity

It's not a docker project for now, but it should be later...

init the project with these commands : this will create a virtual env and install the dependencies

python3 -m venv .venv
source .venv/bin/activate # On windows : .venv\Scripts\activate
make install              # Windows without make? run scripts/install-windows.bat instead

Usage

Unity

Open the project with Unity 2022.3.13f1 or later. You can find the project in the folder UnityProject/.

• Click on the ADD button in the Unity Hub and select the folder UnityProject/
• Select the project and click on the OPEN button

Python

Execute scripts

Open a terminal and run these commands : This will run the script src/pywui/main.py with the --help option to show you the available options

source .venv/bin/activate
pywui --help

Examples

Use your webcam

pywui --source 0

Show the webcam output

pywui --source 0 --show

Use video file

pywui --source path/to/video.mp4

You can also execute the main module directly during development:

python -m pywui.main --help

Execute the tests

Open a terminal and run these commands :

source .venv/bin/activate # On windows : .venv\Scripts\activate
pywui-test

Unity use case

The use case is simple, the python script will send the position of the detected object to Unity. Then Unity will move the object to the position sent by python.

Python

Run the command :

pywui --model ./models/yolov8s-pose.pt --detect-method track --source 0 --conf 0.7 --filter

Unity

Main Scene

• Open the scene Assets/Scenes/Main.unity
• Click on the play button
• You should see as many articulation groups as people detected by the python script and the articulations should move to the position of the people detected
• 

Articulation Scene

• Open the scene Assets/Scenes/Articulation.unity
• Click on the play button
• You should see as many articulation groups as people detected by the python script and the articulations should move to the position of the people detected
• 

Current issues

• python3 might not work, use python instead
• If you have an error with pywui command, try to run source .venv/bin/activate again and then pywui should work
• If you have an error with pywui-test command, try to run source .venv/bin/activate again and then pywui-test should work

Collaborate

If you want to collaborate on this open source repo, you're free to do so.

• Fork the repo
• Create your own branch
• Develop your feature
• And with a PR describe the purpose of your feature