Python code to analyze onnx graph datatypes to solve error "Unsupported ONNX data type: UINT8 (2)"

print_uint_inputs_onnx.py

"""
What:

Upon converting my .onnx model to TensorRT .engine I got this error
Unsupported ONNX data type: UINT8 (2)

So I created this short python function to check which inputs cause the error.
After some googling I found out that uint8 is not supported in TensorRT.

I plan on replacing the uint8's with something else so that I'm able to
convert the model successfully.
"""

import onnx

def main():
    model = onnx.load("onnx_model.onnx")
    try:
        onnx.checker.check_model(model)
    except onnx.checker.ValidationError as e:
        print("The model is invalid: %s" % e)

    """
    # 1 = float32
    # 2 = uint8
    # 3 = int8
    # 4 = uint16
    # 5 = int16
    # 6 = int32
    # 7 = int64
    """
    
    inputs = model.graph.input
    for input in inputs:
        dtype = input.type.tensor_type.elem_type
        if dtype == 2:
            print(input.name, ">>>", input.type.tensor_type.elem_type)

main()

The error may as well result from outputs. You can in theory change the inputs and outputs using this code but that doesn't solve the issue. Basically the whole model must be build without uint8 for it to be compatible with TensorRT