Input shape Error

[Madankumar90]

This is the initial step of the architecture i am using as mentioned in keras2onnx i have put input_shape for the conversion:

class AlexNet(Sequential): def init(self, input_shape, num_classes): super().init()

    self.add(Conv2D(96, kernel_size=(11,11), strides= 4,
                    padding= 'same', use_bias=False,
                    input_shape= input_shape,
                    kernel_initializer= 'he_normal'))
    .....

with the call being: model = AlexNetD((227, 227, 3), num_classes)

and for conversion i am using: onnx_model = keras2onnx.convert_keras(model, model.name)

the conversion is successful but when i run the onnx i am getting this error: OpenCV(4.4.0) Error: Unspecified error (Number of input channels should be multiple of 3 but got 227) in cv::dnn::ConvolutionLayerImpl::getMemoryShapes, file C:\build\master_winpack-build-win64-vc14\opencv\modules\dnn\src\layers\convolution_layer.cpp, line 347 [ERROR:0] global C:\build\master_winpack-build-win64-vc14\opencv\modules\dnn\src\dnn.cpp (3444) cv::dnn::dnn4_v20200609::Net::Impl::getLayerShapesRecursively OPENCV/DNN: [Convolution]:(convolution_output4): getMemoryShapes() throws exception. inputs=1 outputs=0/1 blobs=2 [ERROR:0] global C:\build\master_winpack-build-win64-vc14\opencv\modules\dnn\src\dnn.cpp (3447) cv::dnn::dnn4_v20200609::Net::Impl::getLayerShapesRecursively input[0] = [ 1 227 3 227 ] [ERROR:0] global C:\build\master_winpack-build-win64-vc14\opencv\modules\dnn\src\dnn.cpp (3455) cv::dnn::dnn4_v20200609::Net::Impl::getLayerShapesRecursively blobs[0] = CV_32FC1 [ 96 3 11 11 ]

what am i doing wrong here?? Thanks

[jiafatom]

Looks like your input data dimension does not match the model input dimension. Since the conversion is successful, you can see our test examples to write simple test for the onnx model (running on onnxruntime). See if the simple test can work.

[Madankumar90]

Thanks for the reply . I will check the test examples. should i mention anything about inputs in this onnx_model = keras2onnx.convert_keras(model, model.name) as argument??. My inputs are [2272273]. Thanks.