基于C++OpenVINO部署YOLOX-YOLOV5-YOLOV8-YOLOV9源码.zip

# YOLOX-YOLOV5-YOLOV8-YOLOV9-OpenVINO in C++ This tutorial includes a C++ inference of YOLOX/YOLOV5/YOLOV8/YOLOV9 for OpenVINO. ## Install OpenVINO Toolkit Please visit [Openvino Homepage](https://docs.openvinotoolkit.org/latest/get_started_guides.html) for more details. ## Environment OpenVINO_2022.3.0 OpenCV-4.6.0 ## Convert model 1. Export ONNX model 2. Convert ONNX to OpenVINO Install requirements for convert tool ```shell pip install openvino-dev ``` Convert ONNX into the OpenVINO IR FP32 ```shell mo -m <onnx_model_path> --output_dir <MODEL_DIR> ``` FP16 ```shell mo -m <onnx_model_path> --output_dir <MODEL_DIR> --compress_to_fp16 ``` INT8 Quantization with POT ```shell pot -q default -m <ir_model_xml> -w <ir_model_bin> --engine simplified --data-source <data_dir> --output-dir <output_dir_name> --direct-dump --name <int8_model_name> ``` visit [Openvino POT](https://docs.openvino.ai/latest/notebooks/114-quantization-simplified-mode-with-output.html) for more details. ## Build ### Linux ```shell mkdir build cd build cmake .. make ``` ## Demo ### c++ ```shell ./detect <...><...> ``` ## scrennshot