1.首先安装openvino 2023C++版本，然后将.pth格式的文件转化为.onnx格式文件如最下面的代码块，然后利用openvino的mo模块将.onnx转化为openvino所需的IR格式

mo --input_model <.onnx目录> --output_dir <待保存路径目录>

mo --input_model /home/sunflower/桌面/lens-studio_-yolox-drone_v2.onnx --output_dir /home/sunflower/桌面/xx --compress_to_fp16

import torch
from models import *
from utils import *
cfg_path = '/home/sunflower/belif_plan/反无人机设计/PyTorch-YOLOv3-master/config/yolov3.cfg'
weights_path = '/home/sunflower/belif_plan/反无人机设计/由于嗯/yolov3_ckpt_299.pth'
onnx_path = 'very_good.onnx'
model = Darknet(cfg_path)
model.load_state_dict(torch.load(weights_path, map_location=torch.device('cpu')))
model.eval()
dummy_input = torch.randn(1, 3, 416, 416)
model.to('cpu')
torch.onnx.export(model, dummy_input, onnx_path, input_names=['input'], output_names=['output'],dynamic_axes={'input':{0:'batch_size'},'output':{0:'batch_size'}},opset_version=11)