opencv dnn完整示例

目录

python 导出onnx：

python预测例子：

c++ dnn调用例子：

---

python 导出onnx：

   weights = 'xxxx.pth'

    model_cls= Model_net(model_path=weights, num_classes=2)
    state_dict = torch.load(weights)
    model_cls.load_state_dict(state_dict)
    model_cls.eval()

    dummy_input1 = torch.randn(1, 3, 64, 64)

    onnx_name = "smoke_skipnet.onnx"
    torch.onnx.export(model_cls, dummy_input1, onnx_name, opset_version=11, verbose=False, input_names=['images'], output_names=['output'])

python预测例子：

class Model_net():

    def __init__(self,model_path,num_classes=6):
        self.model = skipnet(num_classes=num_classes)
        self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
        self.num_classes=num_cla