Yolov5 导出ONNX

 1、修改export.py 

      model.model[-1].export = True  
      torch.onnx.export(model, img, f, verbose=False, opset_version=12, input_names=['images'],output_names=['output1', 'output2','output3']) 
 

2、获取训练模型 anchors 
 

 import torch 

from models.experimental import attempt_load 

model = attempt_load('last4.pt', map_location=torch.device('cpu')) 

m = model.module.model[-1] if hasattr(model, 'module') else model.model[-1] 

print(m.anchor_grid) 

3、修改C# Dimensions 
  Dimensions=class数量+5