Tensorflow：SSD_mobilenet_v2_2 中的 TFLiteConverter（已保存模型 -> TFLite）

Question

I have installed tensorflow 2.3.1, my ssd_mobile_net_v2_2 was downloaded from https://tfhub.dev/tensorflow/ssd_mobilenet_v2/2 . I want to convert this model to tf_lite version. My code is:

But error occurs:

tensorflow.lite.python.convert.ConverterError: :0: error: loc("Func/StatefulPartitionedCall/input/_0"): requires all operands and results to have compatible element types :0: note: loc("Func/StatefulPartitionedCall/input/_0"): see current operation: %1 = "tf.Identity"(%arg0) {device = ""} : (tensor<1x320x320x3x!tf.quint8>) -> tensor<1x320x320x3xui8>

it seems I need to set the input-data-type, but I don't know how to do that.

Thanks a lot for your replying.

Answer 1

I looked at this and found this:

import tensorflow as tf

converter = tf.lite.TFLiteConverter.from_saved_model("ssd_mobilenet_v2_2",signature_keys=['serving_default'])
converter.optimizations = [tf.lite.Optimize.DEFAULT]
converter.experimental_new_converter = True
converter.target_spec.supported_ops = [tf.lite.OpsSet.TFLITE_BUILTINS, tf.lite.OpsSet.SELECT_TF_OPS]
model = converter.convert()
open("converted_model.tflite", "wb").write(model)

I'm using tensorflow-nightly I didn't tried with another version

Answer 2

solution from: https://github.com/tensorflow/tensorflow/issues/42114#issuecomment-671593386

note: tf-nightly==2.4