YOLOv5 With OpenCVForUnity Example

• An example of using OpenCV dnn module with YOLOv5. https://github.com/ultralytics/yolov5
• This example can also work for YOLOv6 models, which has the same input/output shapes as the YOLOv5 model.

Environment

• Windows / Mac / Linux / WebGL / Android / iOS
• Unity >= 2021.3.35f1+
• Scripting backend MONO / IL2CPP
• OpenCV for Unity 2.6.4+

Setup

1. Download the latest release unitypackage. YOLOv5WithOpenCVForUnityExample.unitypackage
2. Create a new project. (YOLOv5WithOpenCVForUnityExample)
3. Import OpenCVForUnity.
4. Import the YOLOv5WithOpenCVForUnityExample.unitypackage.
5. Add the "Assets/YOLOv5WithOpenCVForUnityExample/*.unity" files to the "Scenes In Build" list in the "Build Settings" window.
6. Build and Deploy.

Export YOLOv5 model to ONNX

1. YOLOv5_export_to_OpenCVDNN_ONNX.ipynb
2. YOLOv5_segment_export_to_OpenCVDNN_ONNX.ipynb
3. YOLOv5_classify_export_to_OpenCVDNN_ONNX.ipynb

Works with Multi-Object Tracking (MOT)

1. MultiObjectTrackingExample

ScreenShot

Tutorials

1. How to Train YOLO v5 on a Custom Dataset
2. Example of custom training for dice roll detection