quantization of hand_landmark.tflite

[vewe-richard]

I downloaded hand_landmark.tflite from
https://storage.googleapis.com/mediapipe-assets/hand_landmark.tflite?generation=1666153735814956

But hand_landmark.tflite is float 32

I wanted to apply hand_landmark.tflite in nxp imx8 embedded board, but the NPU/GPU only supported int8 and int16.

Can you tell me how to do quantization of hand_landmark.tflite?

From "Model conversion" guide, it seems the conversion need savedModel, but can I get it?

---

The following sections outline the process of evaluating and converting models for use with LiteRT.

Input model formats
You can use the converter with the following input model formats:

[SavedModel]: A TensorFlow model saved as a set of files on disk.
[Keras model]: A model created using the high level Keras API.
[Keras H5 format]: A light-weight alternative to SavedModel format supported by Keras API.
[Models built from concrete functions]: A model created using the low level TensorFlow API.

[vewe-richard]

Hi, @HarishPermal Can you give me some suggestions?

[HarishPermal]

Hi @vewe-richard,

Thank You for being patient with the issue, im currently working on it, and trying out the options available.
I'll update you with the status

Thank You

[HarishPermal]

Hi @schmidt-sebastian Could you please look into this.

Thank You