关于opencv调用onnx模型的一个错误 Can’t infer a dim denoted by -1 in function ‘cv::dnn::computeShapeByReshapeMas

关于opencv调用onnx模型的一个错误

Can’t infer a dim denoted by -1 in function ‘cv::dnn::computeShapeByReshapeMask’

opencv报错：

Traceback (most recent call last):
  File "E:/First_graduate/fatigue/Fatigue_mediapipe0622/Fatigue_Mediapipe0622/pth2onnx_test.py", line 40, in <module>
    outs = net.forward(layerNames)
cv2.error: OpenCV(4.5.5) D:\a\opencv-python\opencv-python\opencv\modules\dnn\src\layers\reshape_layer.cpp:148: error: (-1:Backtrace) Can't infer a dim denoted by -1 in function 'cv::dnn::computeShapeByReshapeMask'

错误大概是网络中的维度问题

错误原因：

输入图像大小与模型的输入大小不一致。
生成onnx时，指定模型输入大小是84x84的

input1 = torch.randn(1, 3, 84, 84)
input_names = ["input"]
output_names = ["output"]
torch.onnx.export(net, input1, "mouthNet2.onnx", verbose=True, input_names=input_names, output_names=output_names)

所以在调用模型推理时输入图像也应当是84x84的。

解决

resize一下即可

img_t = cv.resize(img,(84,84))    #将图片改为模型适用的尺寸