keep_prob=tf. Placeholder (dtype=tf float32, shape=[], name='keep_prob') 
Images=tf. Placeholder (dtype=tf float32, shape=[None, 64, 64, 1), name='image_batch') 
Labels=tf. Placeholder (dtype=tf int64, shape=[None], name='label_batch') 
Is_training=tf. Placeholder (dtype=tf bool, shape=[], name='train_flag') 
With tf. Device ('/CPU: 0 ') : 
With slim. Arg_scope ([slim. Conv2d, slim. Fully_connected], 
Normalizer_fn=slim batch_norm, 
Normalizer_params={' is_training: is_training}) : 
Conv3_1=slim. Conv2d (images, 64, (3, 3), 1, padding='the SAME', the scope='conv3_1') 
Max_pool_1=slim. Max_pool2d (conv3_1 [2, 2], [2, 2], the padding='the SAME', the scope='pool1') 
Conv3_2=slim. Conv2d (max_pool_1, 128, (3, 3), padding='the SAME', the scope='conv3_2') 
Max_pool_2=slim. Max_pool2d (conv3_2 [2, 2], [2, 2], the padding='the SAME', the scope='pool2') 
Conv3_3=slim. Conv2d (max_pool_2, 256, (3, 3), padding='the SAME', the scope='conv3_3') 
Max_pool_3=slim. Max_pool2d (conv3_3 [2, 2], [2, 2], the padding='the SAME', the scope='pool3') 
Conv3_4=slim. Conv2d (max_pool_3, 512, (3, 3), padding='the SAME', the scope='conv3_4') 
Conv3_5=slim. Conv2d (conv3_4, 512, (3, 3), padding='the SAME', the scope='conv3_5') 
Max_pool_4=slim. Max_pool2d (conv3_5 [2, 2], [2, 2], the padding='the SAME', the scope='pool4') 

Flatten=slim. Flatten (max_pool_4) 
Fc1=slim. Fully_connected (slim. Dropout (flatten, keep_prob), 1024, 
Activation_fn=tf. Nn. Relu, scope='fc1') 
Logits=slim. Fully_connected (slim. Dropout (fc1, keep_prob), FLAGS, charset_size, activation_fn=None, 
The scope='fc2) 
Loss=tf. Reduce_mean (tf) nn) sparse_softmax_cross_entropy_with_logits (logits=logits, labels=labels)) 
Accuracy=tf. Reduce_mean (tf) cast (tf) equal (tf) argmax (logits, 1), labels), tf, float32)) 

Update_ops=tf. Get_collection (tf) GraphKeys) update_ops) 
If update_ops: 
Update_ops updates=tf group (*) 
Loss=control_flow_ops. With_dependencies (/updates, loss) 

Global_step=tf. Get_variable (" step ", [],, initializer.=tf constant_initializer (0.0), trainable=False) 
Optimizer=tf. Train. AdamOptimizer (learning_rate=0.1) 
Train_op=slim. Learning. Create_train_op (loss, the optimizer, global_step=global_step) 
Probabilities=tf. Nn. Softmax (logits) 

Tf. The summary. Scalar (' loss 'loss) 
Tf. The summary. Scalar (' accuracy, accuracy) 
Merged_summary_op=tf. The summary. Merge_all () 
Predicted_val_top_k, predicted_index_top_k=tf. Nn. Top_k (probabilities, k=top_k) 
Accuracy_in_top_k=tf. Reduce_mean (tf) cast (tf) nn) in_top_k (probabilities, labels, top_k), tf, float32)) 

 output_node_names="fc2/BatchNorm Reshape_1" 
Saver=tf. Train. Import_meta_graph (input_checkpoint + 'meta', clear_devices=False) 
With tf. The Session () as sess: 
Saver. Restore (sess, input_checkpoint) # recovery figure and get data 
Persistent output_graph_def=graph_util. Convert_variables_to_constants (# model, to fix a variable's value 
Sess=sess, 
Input_graph_def=sess. Graph_def # is equal to: sess. Graph_def 
Output_node_names=output_node_names. Split (", ")) #, if there are multiple output node by commas 

All_output_node_names=[n.n ame for n in output_graph_def. Node] 
Print (STR (all_output_node_names)) 

Pb_file1='./output_graph - 7. Pb '
With tf. Gfile. Gfile (pb_file1, "wb") as f: # save the model 
F.w rite (output_graph_def SerializeToString ()) # serialization output 

 def get_tflite_file (pb_file_path) : 
Input_names=[" image_batch train_flag ", ""," keep_prob "] 
Output_names=[" fc2/BatchNorm Reshape_1 "] 
Input_tensor={" image_batch ": [1, float (64), the float (64), 1]," train_flag ": [True]," keep_prob ": [4]} 
The convert=tf. Lite. TFLiteConverter. From_frozen_graph (pb_file_path input_arrays=input_names, output_arrays=output_names, input_shapes=input_tensor) 
The convert. Post_training_quantize=True 
The convert. Target_spec. Supported_ops=[tf. Lite. OpsSet. TFLITE_BUILTINS, tf, lite. OpsSet. SELECT_TF_OPS] 
The convert. Allow_custom_ops=True 
The convert. Default_ranges_stats=(0, 255) 
Tflite_model=the convert. Convert () 
Open ("../bak/output_graph - 1. Tflite ", "wb"). The write (tflite_model)