Tensorflow "tf.estimator.inputs.numpy_input_fn.numpy_input_fn" not supported by uff conversion

Hello everybody,

I’ve implemented a CNN in Tensorflow (very similar to the CNN MNIST example at https://www.tensorflow.org/tutorials/layers) and now I’m trying to export it in UFF format. I’ve already successfully freeze my trained CNN, but when I try to use uff.from_tensorflow_frozen_model(…) I got a Warning and then an error (see below).

I think it is caused by “tf.estimator.inputs.numpy_input_fn.numpy_input_fn” using an enqueue function which apparently is not supported by the TensorRT UFF conversion.

How can I get this to work maintaining the same functionalities (e.g. different training and prediction batch sizes)?

Thank you,

Elena

Using output node softmax_tensor
Converting to UFF graph
Warning: No conversion function registered for layer: QueueDequeueManyV2 yet.
Converting as custom op QueueDequeueManyV2 fifo_queue_DequeueMany
name: “fifo_queue_DequeueMany”
op: “QueueDequeueManyV2”
input: “enqueue_input/fifo_queue”
input: “fifo_queue_DequeueMany/n”
attr {
key: “_output_shapes”
value {
list {
shape {
dim {
size: 100
}
}
shape {
dim {
size: 100
}
dim {
size: 1081
}
}
shape {
dim {
size: 100
}
}
}
}
}
attr {
key: “component_types”
value {
list {
type: DT_INT64
type: DT_FLOAT
type: DT_INT32
}
}
}
attr {
key: “timeout_ms”
value {
i: -1
}
}

Traceback (most recent call last):
File “create_uff_from_frozen_graph.py”, line 3, in
uff.from_tensorflow_frozen_model("/home/mahi/Desktop/CNN_LIDAR/Frozen_graph/float32/frozen_graph.pb",[“softmax_tensor”])
File “/usr/lib/python2.7/dist-packages/uff/converters/tensorflow/conversion_helpers.py”, line 113, in from_tensorflow_frozen_model
return from_tensorflow(tf_graphdef, output_nodes, **kwargs)
File “/usr/lib/python2.7/dist-packages/uff/converters/tensorflow/conversion_helpers.py”, line 77, in from_tensorflow
name=“main”)
File “/usr/lib/python2.7/dist-packages/uff/converters/tensorflow/converter.py”, line 74, in convert_tf2uff_graph
uff_graph, input_replacements)
File “/usr/lib/python2.7/dist-packages/uff/converters/tensorflow/converter.py”, line 61, in convert_tf2uff_node
op, name, tf_node, inputs, uff_graph, tf_nodes=tf_nodes)
File “/usr/lib/python2.7/dist-packages/uff/converters/tensorflow/converter.py”, line 31, in convert_layer
fields = cls.parse_tf_attrs(tf_node.attr)
File “/usr/lib/python2.7/dist-packages/uff/converters/tensorflow/converter.py”, line 201, in parse_tf_attrs
for key, val in attrs.items()}
File “/usr/lib/python2.7/dist-packages/uff/converters/tensorflow/converter.py”, line 201, in
for key, val in attrs.items()}
File “/usr/lib/python2.7/dist-packages/uff/converters/tensorflow/converter.py”, line 196, in parse_tf_attr_value
return cls.convert_tf2uff_field(code, val)
File “/usr/lib/python2.7/dist-packages/uff/converters/tensorflow/converter.py”, line 182, in convert_tf2uff_field
return uff.List(uff_code, [cls.convert_tf2uff_field(code, v) for v in field_value])
TypeError: ‘ListValue’ object is not iterable