TFLite Failed to Allocate Tensor

Question

I have an exported Frozen Graph .pb file, and converted it to tflite using

graph_def_file = "model.pb"
input_arrays = ["Placeholder"]
input_shape = {"Placeholder": [1024, 2048, 3]}
output_arrays = ["final_output"]
converter = tf.compat.v1.lite.TFLiteConverter.from_frozen_graph(
    graph_def_file, input_arrays, output_arrays, input_shape)

tflite_model = converter.convert()
with open("my_model.tflite", "wb") as f:
        f.write(tflite_model)

And I could not restore my model with

interpreter = tf.lite.Interpreter(model_path="my_model.tflite")
interpreter.resize_tensor_input(0, [1024, 2048, 3], strict=True)
interpreter.allocate_tensors()
---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-28-3cfaee7dc51c> in <module>
      1 interpreter = tf.lite.Interpreter(model_path=model_path)
      2 interpreter.resize_tensor_input(0, [1024, 2048, 3], strict=True)
----> 3 interpreter.allocate_tensors()

~/.local/lib/python3.9/site-packages/tensorflow/lite/python/interpreter.py in allocate_tensors(self)
    512   def allocate_tensors(self):
    513     self._ensure_safe()
--> 514     return self._interpreter.AllocateTensors()
    515 
    516   def _safe_to_run(self):

RuntimeError: tensorflow/lite/kernels/conv.cc:349 input->dims->data[3] != filter->dims->data[3] (65 != 64)Node number 11 (CONV_2D) failed to prepare.Failed to apply the default TensorFlow Lite delegate indexed at 0.

However, I can successfully reload my .pb file frozen graph by using

with tf.gfile.GFile('my_model.pb', "rb") as pb:
    graph_def = tf.GraphDef()
    graph_def.ParseFromString(pb.read())
with tf.Graph().as_default() as graph:
    tf.import_graph_def(
            graph_def,
            name="", 
            )

and also successfully generated result by

node_in = graph.get_tensor_by_name('Placeholder:0') 
node_out = graph.get_tensor_by_name('final_output:0') 

with tf.Session(graph=graph) as sess:  # Session()
    # sess.run(tf.global_variables_initializer()) 
    feed_dict = {node_in: input_img}  
    pred = sess.run(node_out, feed_dict) 
    print(pred)
    sess.close()

I've checked the node 11 of the .tflite file in Netron, but everything seemed fine. Node 11 What could be the problem?