YOLOv5 is a model based on a single-stage object detector, which adopts CSPNet and SPPNet structures and achieves the latest performance in classification and detection tasks. YOLOv5 divides the input image into grids of different sizes and predicts the categories and bounding boxes of all objects present in each grid. This model has fast inference speed and high accuracy, and is suitable for various computer vision tasks such as real-time object detection and face detection.
step1:Install opencv
#instal opencv4.6
wget https://github.com/opencv/opencv/archive/refs/tags/4.6.0.tar.gz
cd opencv
mkdir build
cd build
sudo cmake -D CMAKE_BUILD_TYPE=Release -D CMAKE_INSTALL_PREFIX=/usr/local ..
make
make installstep2:Clone the project and modify the directory where the local ORT is located and compile it
source /opt/dtk/env.sh
cd yolov5_onnxruntime
mkdir build && cd build
cmake ..
make
./YOLOV5
https://github.com/ultralytics/yolov5
https://github.com/itsnine/yolov5-onnxruntime