会话中的输出和feb_dict FailedPreconditionError（请参阅上面的回溯）：尝试使用未初始化的值

Question

我转换MTCNN tensorflow到tensorflow tensorRT

当我运行camera_test.py

我收到此错误FailedPreconditionError: Attempting to use uninitialized in Tensorflow

回溯（最近一次通话）：文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/client/session.py”，第 1334 行，在 _do_call return fn(* args) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotes/lib/python3.6/site-packages/tensorflow/python/client/session.py”，第 1319 行，在 _run_fn 选项、feed_dict、fetch_list、target_list、run_metadata 中）文件“/home/jetsonnano/.virtualenvs/jetsonnanotest/lib/python3.6/site-packages/tensorflow/python/client/session.py”，第1407行，在_call_tf_sessionrun run_metadata中）tensorflow.python.framework.errors_impl.FailedPreconditionError：尝试使用未初始化的值 conv4_2/biases [[{{node conv4_2/biases/read}}]] [[{{node Squeeze_1}}]]

在处理上述异常的过程中，又发生了一个异常：

回溯（最近一次调用）：文件“camera_test_trrt.py”，第 48 行，在 box_c,landmarks = mtcnn_detector.detect(image) 文件“../Detection/MtcnnDetector.py”，第 371 行，在检测盒中，boxes_c， _ = self.detect_pnet(img) File "../Detection/MtcnnDetector.py", line 221, in detect_pnet cls_cls_map, reg = self.pnet_detector.predict(im_resized) File "../Detection/fcn_detector_trrt.py", line 56, in predict self.height_op: height}) 文件 "/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/client/session.py", line 929, in run_metadata_ptr ) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/client/session.py”，第 1152 行，在 _run feed_dict_tensor, options, run_metadata) 文件“/home/ jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/client/session.py", line 1328, in _do_run run_metadata) 文件"/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3 .6/site-packages/tensorf low/python/client/session.py”，第 1348 行，在 _do_call raise type(e)(node_def, op, message) tensorflow.python.framework.errors_impl.FailedPreconditionError: Attempting to use uninitialized value conv4_2/biases [[node conv4_2 /biases/read（定义在 ../train_models/mtcnn_model.py:208）]] [[node Squeeze_1（定义在 ../train_models/mtcnn_model.py:245）]]

由 op 'conv4_2/biases/read' 引起，定义在：文件“camera_test_trrt.py”，第 23 行，在 PNet = FcnDetector(P_Net, '/home/jetsonnano/Downloads/MTCNN-Tensorflow-master/test/p_output_graph_FP16.pb ') 文件“../Detection/fcn_detector_trrt.py”，第 23 行，在init self.cls_prob, self.bbox_pred, _ = net_factory(image_reshape, training=False) 文件“../train_models/mtcnn_model.py”，行208、在P_Net中 bbox_pred = slim.conv2d(net,num_outputs=4,kernel_size=[1,1],stride=1,scope='conv4_2',activation_fn=None) 文件"/home/jetsonnano/.virtualenvs/jetsonnanotest/ lib/python3.6/site-packages/tensorflow/contrib/framework/python/ops/arg_scope.py", line 182, in func_with_args return func(*args, **current_args) File "/home/jetsonnano/.virtualenvs/ jetsonnanotetest/lib/python3.6/site-packages/tensorflow/contrib/layers/python/layers/layers.py”，第1158行，在convolution2d conv_dims=2）文件“/home/jetsonnano/.virtualenvs/jetsonnanotest/lib/ python3.6/site-packages/tensorflow/contrib/framework/python/ops/arg_scope.py", l ine 182, in func_with_args return func(*args, **current_args) File "/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/contrib/layers/python/layers/layers.py "，第 1061 行，在卷积输出 = layer.apply(inputs) 文件"/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/keras/engine/base_layer.py"，第 1227 行，在申请中返回自我。 call (inputs, *args, **kwargs) File "/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/layers/base.py", line 530, in call output = 超级（层，自我）。 调用（输入，*args，**kwargs）文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/keras/engine/base_layer.py”，第538行，在调用self._maybe_build(inputs) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotes/lib/python3.6/site-packages/tensorflow/python/keras/engine/base_layer.py”，第 1603 行，在 _maybe_build self.build （input_shapes）文件“/home/jetsonnano/.virtualenvs/jetsonnanotes/lib/python3.6/site-packages/tensorflow/python/keras/layers/convolutional.py”，第174行，在构建dtype=self.dtype）文件“/home/jetsonnano/.virtualenvs/jetsonnanotest/lib/python3.6/site-packages/tensorflow/python/layers/base.py”，第 435 行，在 add_weight getter=vs.get_variable) 文件“/home/jetsonnano/ .virtualenvs/jetsonnanotest/lib/python3.6/site-packages/tensorflow/python/keras/engine/base_layer.py”，第 349 行，在 add_weight 聚合=聚合中）文件“/home/jetsonnano/.virtualenvs/jetsonnanotest/lib /python3.6/site-packages/tensorflow/python/training/checkpointab le/base.py”，第 607 行，在 _add_variable_with_custom_getter **kwargs_for_getter）文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/ops/variable_scope.py”，行1479，在 get_variable 聚合 = 聚合中）文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/ops/variable_scope.py”，第 1220 行，在 get_variable 聚合 = 聚合中）文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/ops/variable_scope.py”，第530行，在get_variable中返回custom_getter(**custom_getter_kwargs)文件“/home/ jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/contrib/layers/python/layers/layers.py”，第 1753 行，在 layer_variable_getter 中返回 _model_variable_getter(getter, *args, **kwargs) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/contrib/layers/python/layers/layers.py”，第 1744 行，在 _model_variable 中 _getter 聚合 = 聚合）文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/contrib/framework/python/ops/arg_scope.py”，第 182 行，在 func_with_args 中 return func( *args, **current_args) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/contrib/framework/python/ops/variables.py”，第350行，在model_variable聚合中=aggregation) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotes/lib/python3.6/site-packages/tensorflow/contrib/framework/python/ops/arg_scope.py”，第 182 行，在 func_with_args 中 return func(*args , **current_args) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/contrib/framework/python/ops/variables.py”，第277行，变量aggregation=aggregation ) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/ops/variable_scope.py”，第 499 行，在 _true_getter 聚合=聚合中) 文件“/home/jetsonnano/ .virtualen vs/jetsonnanotest/lib/python3.6/site-packages/tensorflow/python/ops/variable_scope.py”，第 911 行，在 _get_single_variable 聚合=聚合中）文件“/home/jetsonnano/.virtualenvs/jetsonnanotest/lib/python3. 6/site-packages/tensorflow/python/ops/variables.py”，第 213 行，在调用return cls._variable_v1_call(*args, **kwargs) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotest/lib/python3. 6/site-packages/tensorflow/python/ops/variables.py”，第 176 行，在 _variable_v1_call 聚合=聚合中）文件“/home/jetsonnano/.virtualenvs/jetsonnanotes/lib/python3.6/site-packages/tensorflow/ python/ops/variables.py", line 155, in previous_getter = lambda **kwargs: default_variable_creator(None, **kwargs) File "/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/ tensorflow/python/ops/variable_scope.py", line 2495, in default_variable_creator expected_shape=expected_shape, import_scope=import_scope) 文件 "/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages /tensorflow/python/ops/variables.py”，第 217 行，在调用return super(VariableMetaclass, cls) 中。 调用(*args, **kwargs) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/ops/variables.py”，第 1395 行，在init constraint=constraint ) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/ops/variables.py”，第 1557 行，在 _init_from_args self._snapshot = array_ops.identity(self._variable , name="read") 文件“/home/jetsonnano/.virtualenvs/jetsonnanotes/lib/python3.6/site-packages/tensorflow/python/util/dispatch.py​​”，第180行，包装器返回目标（*args , **kwargs) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/ops/array_ops.py”，第 81 行，身份为 ret = gen_array_ops.identity(input , name=name) 文件“/home/jetsonnano/.virtualenvs/jetsonnanotes/lib/python3.6/site-packages/tensorflow/python/ops/gen_array_ops.py”，第3890行，身份“Identity”，输入=输入, name=name) 文件 "/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-package s/tensorflow/python/framework/op_def_library.py”，第 788 行，在 _apply_op_helper op_def=op_def）文件“/home/jetsonnano/.virtualenvs/jetsonnanotest/lib/python3.6/site-packages/tensorflow/python/util/ deprecation.py", line 507, in new_func return func(*args, **kwargs) File "/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/framework/ops. py", line 3300, in create_op op_def=op_def) 文件 "/home/jetsonnano/.virtualenvs/jetsonnanotetest/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 1801, in init self._traceback = tf_stack.extract_stack()

FailedPreconditionError（回溯见上文）：尝试使用未初始化的值 conv4_2/biases [[node conv4_2/biases/read（定义于 ../train_models/mtcnn_model.py:208）]] [[node Squeeze_1（定义于 ../ train_models/mtcnn_model.py:245)]]

我如何tf.global_variables_initializer将sess.run

init_op = tf.initialize_all_variables()

sess = tf.Session()
sess.run(init_op)

当我有output parameters和feed_dict在sess.run

cls_prob，bbox_pred，landmark_pred = self.sess.run（[self.cls_prob，self.bbox_pred，self.landmark_pred]，feed_dict = {self.image_op：数据}）在detector.py

和

cls_prob, bbox_pred = self.sess.run([self.cls_prob, self.bbox_pred],feed_dict={self.image_op: databatch, self.width_op: width,self.height_op: height}) in fcn_detector.py

有人可以帮忙吗？

Answer 1

就在以下行之后

self.sess = tf.Session( config=tf.ConfigProto(allow_soft_placement=True, gpu_options=tf.GPUOptions(allow_growth=True)))

宣布

init_op = tf.global_variables_initializer() 

并做

self.sess.run(init_op)