Currently if you try and run the TensorFlowSharp library against the model it will request a different image format