Hello,
I am trying to run object detection script from below link using saved trained model.
link - Retraining an Image Classifier | TensorFlow Hub
I have used “efficientnetv2-xl-21k-ft1k” pre-trained model for training.
object detection script not allowing to use image with resolution other than 512*512 and datatype other that float32 while reshaping.
Is there any way to use image with any resolution as well as any of the data type?
I got below error :
Code :-
running inference
results = hub_model(image_np)
print(results)
result = {key:value.numpy() for key,value in results.items()}
print(result.keys())
Error : -
ValueError Traceback (most recent call last)
in <cell line: 2>()
1 # running inference
----> 2 results = hub_model(image_np)
3 print(results)
4
5 # different object detection models have additional results
2 frames
/usr/local/lib/python3.10/dist-packages/tensorflow/python/saved_model/function_deserialization.py in restored_function_body(*args, **kwargs)
333 “Option {}:\n {}\n Keyword arguments: {}”.format(
334 index + 1, _pretty_format_positional(positional), keyword))
→ 335 raise ValueError(
336 "Could not find matching concrete function to call loaded from the "
337 f"SavedModel. Got:\n {_pretty_format_positional(args)}\n Keyword "
ValueError: Could not find matching concrete function to call loaded from the SavedModel. Got:
Positional arguments (3 total):
* <tf.Tensor ‘inputs:0’ shape=(1, 2769, 1698, 3) dtype=uint8>
* False
* None
Keyword arguments: {}
Expected these arguments to match one of the following 2 option(s):
Option 1:
Positional arguments (3 total):
* TensorSpec(shape=(None, 512, 512, 3), dtype=tf.float32, name=‘input_1’)
* True
* None
Keyword arguments: {}
Option 2:
Positional arguments (3 total):
* TensorSpec(shape=(None, 512, 512, 3), dtype=tf.float32, name=‘input_1’)
* False
* None
Keyword arguments: {}