Convert a tensorflow detection model using coremltools

I have created a detection model using this tutorial. https://tensorflow-object-detection-api-tutorial.readthedocs.io/en/latest/training.html#

Can I convert it using coreml tools ?

When I use tf.saved_model.load I get a

<tensorflow.python.saved_model.load.Loader._recreate_base_user_object.<locals>._UserObject object at 0x000001C86252F3D0>.

And thus coremltools return this error.

NotImplementedError: Expected model format: [SavedModel | [concrete_function] | tf.keras.Model | .h5], got <tensorflow.python.saved_model.load.Loader._recreate_base_user_object.._UserObject object at 0x000001C86252F3D0>

What is this object ?

Thank you !

code sample :

new_model = tf.saved_model.load("my-model/saved_model")
image_input = ct.ImageType(shape=(1, 224, 224, 3,),
                          bias=[-1,-1,-1], scale=1/127)
core_ml_model = ct.convert( new_model, inputs=[image_input], source="tensorflow")
core_ml_model.save("SSDMobileNetV2.mlmodel")

ps : I also tried core_ml_model = ct.convert("my-model/saved_model") Which give

TypeError: 'Var' object is not subscriptable