这篇文章主要讲解了keras模型如何保存为tensorflow的二进制模型，内容清晰明了，对此有兴趣的小伙伴可以学习一下，相信大家阅读完之后会有帮助。

最近需要将使用keras训练的模型移植到手机上使用， 因此需要转换到tensorflow的二进制模型。

折腾一下午，终于找到一个合适的方法，废话不多说，直接上代码：

# coding=utf-8import sysfrom keras.models import load_modelimport tensorflow as tfimport osimport os.path as ospfrom keras import backend as Kdef freeze_session(session, keep_var_names=None, output_names=None, clear_devices=True): """ Freezes the state of a session into a prunned computation graph. Creates a new computation graph where variable nodes are replaced by constants taking their current value in the session. The new graph will be prunned so subgraphs that are not neccesary to compute the requested outputs are removed. @param session The TensorFlow session to be frozen. @param keep_var_names A list of variable names that should not be frozen, or None to freeze all the variables in the graph. @param output_names Names of the relevant graph outputs. @param clear_devices Remove the device directives from the graph for better portability. @return The frozen graph definition. """ from tensorflow.python.framework.graph_util import convert_variables_to_constants graph = session.graph with graph.as_default(): freeze_var_names = list(set(v.op.name for v in tf.global_variables()).difference(keep_var_names or [])) output_names = output_names or [] output_names += [v.op.name for v in tf.global_variables()] input_graph_def = graph.as_graph_def() if clear_devices: for node in input_graph_def.node: node.device = "" frozen_graph = convert_variables_to_constants(session, input_graph_def, output_names, freeze_var_names) return frozen_graphinput_fld = sys.path[0]weight_file = 'your_model.h6'output_graph_name = 'tensor_model.pb'output_fld = input_fld + '/tensorflow_model/'if not os.path.isdir(output_fld): os.mkdir(output_fld)weight_file_path = osp.join(input_fld, weight_file)K.set_learning_phase(0)net_model = load_model(weight_file_path)print('input is :', net_model.input.name)print ('output is:', net_model.output.name)sess = K.get_session()frozen_graph = freeze_session(K.get_session(), output_names=[net_model.output.op.name])from tensorflow.python.framework import graph_iograph_io.write_graph(frozen_graph, output_fld, output_graph_name, as_text=False)print('saved the constant graph (ready for inference) at: ', osp.join(output_fld, output_graph_name))

上面代码实现保存到当前目录的tensor_model目录下。

验证：

import tensorflow as tfimport numpy as npimport PIL.Image as Imageimport cv2def recognize(jpg_path, pb_file_path): with tf.Graph().as_default(): output_graph_def = tf.GraphDef() with open(pb_file_path, "rb") as f: output_graph_def.ParseFromString(f.read()) tensors = tf.import_graph_def(output_graph_def, name="") print tensors with tf.Session() as sess: init = tf.global_variables_initializer() sess.run(init) op = sess.graph.get_operations() for m in op: print(m.values()) input_x = sess.graph.get_tensor_by_name("convolution2d_1_input:0") #具体名称看上一段代码的input.name print input_x out_softmax = sess.graph.get_tensor_by_name("activation_4/Softmax:0") #具体名称看上一段代码的output.name print out_softmax img = cv2.imread(jpg_path, 0) img_out_softmax = sess.run(out_softmax, feed_dict={input_x: 1.0 - np.array(img).reshape((-1,28, 28, 1)) / 255.0}) print "img_out_softmax:", img_out_softmax prediction_labels = np.argmax(img_out_softmax, axis=1) print "label:", prediction_labelspb_path = 'tensorflow_model/constant_graph_weights.pb'img = 'test/6/8_48.jpg'recognize(img, pb_path)

补充知识：如何将keras训练好的模型转换成tensorflow的.pb的文件并在TensorFlow serving环境调用

首先keras训练好的模型通过自带的model.save()保存下来是 .model （.h6) 格式的文件

模型载入是通过 my_model = keras . models . load_model( filepath )

要将该模型转换为.pb 格式的TensorFlow 模型，代码如下：

# -*- coding: utf-8 -*-from keras.layers.core import Activation, Dense, Flattenfrom keras.layers.embeddings import Embeddingfrom keras.layers.recurrent import LSTMfrom keras.layers import Dropoutfrom keras.layers.wrappers import Bidirectionalfrom keras.models import Sequential,load_modelfrom keras.preprocessing import sequencefrom sklearn.model_selection import train_test_splitimport collectionsfrom collections import defaultdictimport jiebaimport numpy as npimport sysreload(sys)sys.setdefaultencoding('utf-8')import tensorflow as tfimport osimport os.path as ospfrom keras import backend as Kdef freeze_session(session, keep_var_names=None, output_names=None, clear_devices=True): from tensorflow.python.framework.graph_util import convert_variables_to_constants graph = session.graph with graph.as_default(): freeze_var_names = list(set(v.op.name for v in tf.global_variables()).difference(keep_var_names or [])) output_names = output_names or [] output_names += [v.op.name for v in tf.global_variables()] input_graph_def = graph.as_graph_def() if clear_devices: for node in input_graph_def.node: node.device = "" frozen_graph = convert_variables_to_constants(session, input_graph_def, output_names, freeze_var_names) return frozen_graphinput_fld = '/data/codebase/Keyword-fenci/brand_recogniton_biLSTM/'weight_file = 'biLSTM_brand_recognize.model'output_graph_name = 'tensor_model_v3.pb'output_fld = input_fld + '/tensorflow_model/'if not os.path.isdir(output_fld): os.mkdir(output_fld)weight_file_path = osp.join(input_fld, weight_file)K.set_learning_phase(0)net_model = load_model(weight_file_path)print('input is :', net_model.input.name)print ('output is:', net_model.output.name)sess = K.get_session()frozen_graph = freeze_session(K.get_session(), output_names=[net_model.output.op.name])from tensorflow.python.framework import graph_iograph_io.write_graph(frozen_graph, output_fld, output_graph_name, as_text=True)print('saved the constant graph (ready for inference) at: ', osp.join(output_fld, output_graph_name))

然后模型就存成了.pb格式的文件

问题就来了，这样存下来的.pb格式的文件是frozen model

如果通过TensorFlow serving 启用模型的话，会报错：

E tensorflow_serving/core/aspired_versions_manager.cc:358] Servable {name: mnist version: 1} cannot be loaded: Not found: Could not find meta graph def matching supplied tags: { serve }. To inspect available tag-sets in the SavedModel, please use the SavedModel CLI: `saved_model_cli`

因为TensorFlow serving 希望读取的是saved model

于是需要将frozen model 转化为 saved model 格式，解决方案如下:

from tensorflow.python.saved_model import signature_constantsfrom tensorflow.python.saved_model import tag_constantsexport_dir = '/data/codebase/Keyword-fenci/brand_recogniton_biLSTM/saved_model'graph_pb = '/data/codebase/Keyword-fenci/brand_recogniton_biLSTM/tensorflow_model/tensor_model.pb'builder = tf.saved_model.builder.SavedModelBuilder(export_dir)with tf.gfile.GFile(graph_pb, "rb") as f: graph_def = tf.GraphDef() graph_def.ParseFromString(f.read())sigs = {}with tf.Session(graph=tf.Graph()) as sess: # name="" is important to ensure we don't get spurious prefixing tf.import_graph_def(graph_def, name="") g = tf.get_default_graph() inp = g.get_tensor_by_name(net_model.input.name) out = g.get_tensor_by_name(net_model.output.name) sigs[signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY] = \ tf.saved_model.signature_def_utils.predict_signature_def( {"in": inp}, {"out": out}) builder.add_meta_graph_and_variables(sess, [tag_constants.SERVING], signature_def_map=sigs)builder.save()

于是保存下来的saved model 文件夹下就有两个文件：

saved_model.pb variables

其中variables 可以为空

于是将.pb 模型导入serving再读取，成功！

看完上述内容，是不是对keras模型如何保存为tensorflow的二进制模型有进一步的了解，如果还想学习更多内容，欢迎关注亿速云行业资讯频道。