In older versions of keras (such as 2.2.4), conv_utils could be imported with
from keras.utils import conv_utils
After trying to find out how to import this module in newer versions, an older question helped me to import it as
import tensorflow as tf
import tensorflow.keras.utils as conv_utils
But there are some functions from the older conv_utils that cannot be used from tensorflow.keras.utils such as deconv_length or normalize_tuple, and if I run dir(conv_utils) these functions are not there:
>>>dir(conv_utils)
['CustomObjectScope',
'GeneratorEnqueuer',
'OrderedEnqueuer',
'Progbar',
'Sequence',
'SequenceEnqueuer',
'__builtins__',
'__cached__',
'__doc__',
'__file__',
'__loader__',
'__name__',
'__package__',
'__path__',
'__spec__',
'_sys',
'custom_object_scope',
'deserialize_keras_object',
'get_custom_objects',
'get_file',
'get_registered_name',
'get_registered_object',
'get_source_inputs',
'model_to_dot',
'normalize',
'pack_x_y_sample_weight',
'plot_model',
'register_keras_serializable',
'serialize_keras_object',
'to_categorical',
'unpack_x_y_sample_weight']