YOLOv5 Inference C++

This example demonstrates how to perform inference using YOLOv5 models in C++ with OpenCV's DNN API.

This repository was originally created by Matin Ghorbani and is available at this original repository

Usage

Clone the repository

git clone https://github.com/matin-ghorbani/Yolov5-CPP.git

Copy the onnx models in the models directory

First you need to download the weight that you want from here

Then you need to convert the .pt format to .onnx format To export YOLOv5 models:

First install ultralytics

pip install ultralytics

Then clone the ultralytics repository and convert the .pt format to .onnx format with this command:

git clone https://github.com/ultralytics/yolov5.git
cd yolov5
python3 export.py --weights YOUR_WEIGHT.pt --img 480 640 --include onnx --opset 12

Finally, Copy the YOUR_WEIGHT.onnx file to assets/models directory.

Go to the repository directory and make a build directory

cd Yolov5-CPP/
mkdir build
cd build/

Config CMake

cmake ..

Build

make

Run and detect:

• On Image: ./run YOUR_IMG.JPG
• On Video: ./run YOUR_VIDEO.MP4
• On Webcam: ./run YOUR_WEBCAM_ID

Result

Note that the example networks are exported with rectangular (640x480) resolutions