Loading model fails

[dohtem81]

Running on MacOS, Docker Desktop 2.3.0.3, python:3.7 image. Installed all dependencies. When loading model resnet50_coco_best_v2.1.0.h5 following log ends up with error

Code pretty straightforward:

from keras_retinanet.models import load_model model = load_model('./models/resnet50_coco_best_v2.1.0.h5', backbone_name='resnet50')

Using TensorFlow backend.
2020-06-23 03:53:58.307402: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcuda.so.1'; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory
2020-06-23 03:53:58.307474: E tensorflow/stream_executor/cuda/cuda_driver.cc:313] failed call to cuInit: UNKNOWN ERROR (303)
2020-06-23 03:53:58.307510: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (c701da7b5de1): /proc/driver/nvidia/version does not exist
2020-06-23 03:53:58.307683: I tensorflow/core/platform/cpu_feature_guard.cc:143] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2020-06-23 03:53:58.313914: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2592000000 Hz
2020-06-23 03:53:58.314329: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x7f1404000b20 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-06-23 03:53:58.314386: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
Traceback (most recent call last):
  File "conertmodel.py", line 2, in <module>
    model = load_model('./models/resnet50_coco_best_v2.1.0.h5', backbone_name='resnet50')
  File "/usr/local/lib/python3.7/site-packages/keras_retinanet/models/__init__.py", line 87, in load_model
    return keras.models.load_model(filepath, custom_objects=backbone(backbone_name).custom_objects)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/saving/save.py", line 184, in load_model
    return hdf5_format.load_model_from_hdf5(filepath, custom_objects, compile)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/saving/hdf5_format.py", line 178, in load_model_from_hdf5
    custom_objects=custom_objects)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/saving/model_config.py", line 55, in model_from_config
    return deserialize(config, custom_objects=custom_objects)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/layers/serialization.py", line 109, in deserialize
    printable_module_name='layer')
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/utils/generic_utils.py", line 373, in deserialize_keras_object
    list(custom_objects.items())))
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/engine/network.py", line 987, in from_config
    config, custom_objects)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/engine/network.py", line 2019, in reconstruct_from_config
    process_layer(layer_data)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/engine/network.py", line 2001, in process_layer
    layer = deserialize_layer(layer_data, custom_objects=custom_objects)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/layers/serialization.py", line 109, in deserialize
    printable_module_name='layer')
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/utils/generic_utils.py", line 373, in deserialize_keras_object
    list(custom_objects.items())))
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/engine/network.py", line 987, in from_config
    config, custom_objects)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/engine/network.py", line 2029, in reconstruct_from_config
    process_node(layer, node_data)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/engine/network.py", line 1977, in process_node
    output_tensors = layer(input_tensors, **kwargs)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/engine/base_layer.py", line 897, in __call__
    self._maybe_build(inputs)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/engine/base_layer.py", line 2416, in _maybe_build
    self.build(input_shapes)  # pylint:disable=not-callable
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/layers/convolutional.py", line 172, in build
    dtype=self.dtype)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/engine/base_layer.py", line 577, in add_weight
    caching_device=caching_device)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/training/tracking/base.py", line 743, in _add_variable_with_custom_getter
    **kwargs_for_getter)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/engine/base_layer_utils.py", line 141, in make_variable
    shape=variable_shape if variable_shape else None)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/ops/variables.py", line 259, in __call__
    return cls._variable_v1_call(*args, **kwargs)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/ops/variables.py", line 220, in _variable_v1_call
    shape=shape)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/ops/variables.py", line 198, in <lambda>
    previous_getter = lambda **kwargs: default_variable_creator(None, **kwargs)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/ops/variable_scope.py", line 2598, in default_variable_creator
    shape=shape)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/ops/variables.py", line 263, in __call__
    return super(VariableMetaclass, cls).__call__(*args, **kwargs)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/ops/resource_variable_ops.py", line 1434, in __init__
    distribute_strategy=distribute_strategy)
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/ops/resource_variable_ops.py", line 1567, in _init_from_args
    initial_value() if init_from_fn else initial_value,
  File "/usr/local/lib/python3.7/site-packages/tensorflow/python/keras/engine/base_layer_utils.py", line 121, in <lambda>
    init_val = lambda: initializer(shape, dtype=dtype)
  File "/usr/local/lib/python3.7/site-packages/keras_retinanet/initializers.py", line 37, in __call__
    result = np.ones(shape, dtype=dtype) * -math.log((1 - self.probability) / self.probability)
  File "/usr/local/lib/python3.7/site-packages/numpy/core/numeric.py", line 192, in ones
    a = empty(shape, dtype, order)
TypeError: Cannot interpret 'tf.float32' as a data type

[ptdw]

I'm having the same issue with python3.8.

[shivareddy37]

having same issue with python 3.7 keras=2.4.2 and tensorflow=2.2.0

[kkrunal77]

having same issue with python 3.7 keras=2.4.3 and tensorflow=2.2.0

[eeilerts]

I had the same problem. Turns out that 8 days ago Keras updated to version 2.4 with a note saying that some workflows may break. I rolled back to keras 2.3.1 and everything worked again:

pip3 install keras==2.3.1

[ptdw]

Thanks @eeilerts, that solved it for me!

[dohtem81]

Looks like this fixed my problem also. Thanks!

On Jun 25, 2020, at 9:00 PM, ptdw notifications@github.com wrote:

Thanks @eeilerts https://github.com/eeilerts, that solved it for me!

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/fizyr/keras-retinanet/issues/1395#issuecomment-649906366, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACETMVYEJ4EXQAO4GFPBNBLRYP6MFANCNFSM4OFIG2VQ.

[lulmer]

Hello, I have the same problem, I replaced all imports by tf.keras (version id 2.3.0-tf) Are tf.keras and keras incompatible ? Thank you

[dohtem81]

No. All is the same. Just use Keras 2.3.1 instead of latest. I got caught by that because I started to work on something, got a new laptop and when moving source I did not think that Keras version between laptops is not the same. Shame on me ;)

On Tue, Jul 7, 2020 at 7:50 AM Louis Ulmer notifications@github.com wrote:

Hello, I have the same problem, I replaced all imports by tf.keras (version id 2.3.0-tf) Are tf.keras and keras incompatible ? Thank you

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/fizyr/keras-retinanet/issues/1395#issuecomment-654834309, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACETMV7WRM2CTAKC45I54RTR2MKZTANCNFSM4OFIG2VQ .

[Prashantmdgl9]

Downgrading Keras to 2.3.1 works.

[lulmer]

Indeed, downgrading keras to 2.3.1 works for me as well. Thanks

[mooratov]

is it worth pinning setup.py to Keras 2.3.1? (This is what I have now done as a temporary hotfix on private fork). Otherwise, seems there is a good bit of work to do to get various things compatible with 2.4

[hgaiser]

What is the issue with 2.4?

[dohtem81]

Cannot load model. Looks like this is the final effect but the root cause seems to be in new Keras. TypeError: Cannot interpret 'tf.float32' as a data type

On Thu, Jul 23, 2020 at 2:03 AM Hans Gaiser notifications@github.com wrote:

What is the issue with 2.4?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/fizyr/keras-retinanet/issues/1395#issuecomment-662850771, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACETMV2PXACT7J4WTTR6BSDR47OFVANCNFSM4OFIG2VQ .

[jhoncc2]

Hi, I have a similar issue. I'm using Google Colab, Python 3.6.9, tensorflow 2.2

TypeError                                 Traceback (most recent call last)
<ipython-input-3-7ddd8b74fd4e> in <module>()
      5 # load retinanet model
      6 model_path = os.path.join( 'snapshots', 'resnet50_coco_best_v2.1.0.h5')
----> 7 model = models.load_model(model_path, backbone_name='resnet50')
      8 
      9 # if the model is not converted to an inference model, use the line below

31 frames
/usr/local/lib/python3.6/dist-packages/numpy/core/numeric.py in ones(shape, dtype, order)
    205 
    206     """
--> 207     a = empty(shape, dtype, order)
    208     multiarray.copyto(a, 1, casting='unsafe')
    209     return a

TypeError: data type not understood

after changing Keras from 2.4.3 to 2.3.1 I get following error

AttributeError                            Traceback (most recent call last)
<ipython-input-6-7ddd8b74fd4e> in <module>()
      5 # load retinanet model
      6 model_path = os.path.join( 'snapshots', 'resnet50_coco_best_v2.1.0.h5')
----> 7 model = models.load_model(model_path, backbone_name='resnet50')
      8 
      9 # if the model is not converted to an inference model, use the line below

/usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py in is_tensor(x)
    701 
    702 def is_tensor(x):
--> 703     return isinstance(x, tf_ops._TensorLike) or tf_ops.is_dense_tensor
    704 
    705 

AttributeError: module 'tensorflow.python.framework.ops' has no attribute '_TensorLike'

[jhoncc2]

Seems that Colab has changed the versions of Keras and TF. This problems are solved downgrading Keras and TF to versions Keras=2.3.1 and TF=2.1. Using newer versions of Keras and Tensorflow make this problem appear. Hope it helps.

[CGuerrero13]

Seems that Colab has changed the versions of Keras and TF. This problem does not appear versions Keras 2.3.1, and TF=2.1. I tested in newer versions of Keras and Tensorflow, and the problem persist. Hope it helps

I have tested with keras 2.3.1 and tf 2.2 in Colab and does not work. I changed the versions in requirements.txt file and run it.

[jhoncc2]

Seems that Colab has changed the versions of Keras and TF. This problem does not appear versions Keras 2.3.1, and TF=2.1. I tested in newer versions of Keras and Tensorflow, and the problem persist. Hope it helps

I have tested with keras 2.3.1 and tf 2.2 in Colab and does not work. I changed the versions in requirements.txt file and run it.

Try downgrading Tf=2.1, it solved this problem to me. i changed my original comment to make it more understandable :)

[CGuerrero13]

Thank you @jhoncc2 . Colab works with new tensorflow versions. Now the code works!

[sri-dhurkesh]

Seems that Colab has changed the versions of Keras and TF. This problem does not appear versions Keras 2.3.1, and TF=2.1. I tested in newer versions of Keras and Tensorflow, and the problem persist. Hope it helps

I have tested with keras 2.3.1 and tf 2.2 in Colab and does not work. I changed the versions in requirements.txt file and run it.

Try downgrading Tf=2.1, it solved this problem to me. i changed my original comment to make it more understandable :)

thanks alot @jhoncc2

[stale[bot]]

This issue has been automatically marked as stale due to the lack of recent activity. It will be closed if no further activity occurs. Thank you for your contributions.