YOLOv5-Openvino-Cpp-Python

Example of performing inference with ultralytics YOLOv5 using the 2022.1.0 openvino API in C++ using Docker as well as python.

This repository is only for model inference using openvino. Therefore, it assumes the YOLOv5 model is already trained and exported to openvino (.bin, .xml) format. For further info check YOLOv5.

YOLOv5-Openvino-Cpp

Docker installation

This repository folder contains the Dockerfile to build a docker image with the Intel® Distribution of OpenVINO™ toolkit.

1. This command builds an image with OpenVINO™ 2022.1.0 release.

   docker build cpp -t openvino_container:2022.1.0
   

2. This command creates a docker container with OpenVINO™ 2022.1.0 release.

   Windows

   docker run -it --rm -v %cd%:/yolov5-openvino openvino_container:2022.1.0
   

   Linux/Mac

   docker run -it --rm -v $(pwd):/yolov5-openvino openvino_container:2022.1.0
   

Cmake build

From within the docker run:

cd cpp && mkdir build && cd build

Then create the make file using cmake:

cmake -S ../ -O ./

Then compile the program using:

make

Then run the executable:

./main

YOLOv5-Openvino-Python

Usage

python -m venv /path/to/env

source /path/to/env/bin/activate # Linux/mac

\path\to\env\Script\activate # Windows

cd python

pip install -r requirements.txt

Then run the script:

python main.py

Final Result: