Part Number: TDA4VM
Tool/software:
I tried compiling my model with both 16 and 8 bit but the compiled model that I see from SVG file generated is different. Furthermore the bin file generated is in accordance to the SVG file rather than the expected output of the onnx model.
I am supposed to have 3 outputs: Masks(coming as expected), 2 concat layers each concatenating 3 different inputs.
So in the concat layer, the concat operator is taking 2 inputs as expected but for the third input it is just flattening 1 input and taking it as the 3rd input. The 2 inputs that should've been there in place of the flattened input are concatenated as a 4th output.
I am unable to upload the full SVG so I have uploaded one of the relevant layers to this. The other concat layer has the same issue. I am using SDK 8.2.

