Tensorflow_Mobilenet_transfer_learning

transfor learning of Mobilenet by tensorflow,include train , test, frozen graph

#five script as follows:

	'''
	run_mobilenet.sh
	run_inception.sh
	run_evaluate.sh
	run_frozen_graph.sh
	run_count_ops.sh
	'''

1,train or test or optimize or quantize or tflite for mobilenet/inception.When setting the checkpoint_path,it could load latest checkpoint and restore values,and train the model by this weight values.If you set output_graph,you will get a pb after retrain finishs.

	run_mobilenet.sh
	run_inception.sh

2,run_evaluate.sh——eval the model and print the final accuracy

3,run_frozen_graph.sh——Genetare the pb file through the ckpt, read this pb file and print operation. That's important,you must set the correct input_names and output_names.If you do'nt know the name of input and output layer,you coud run the follow function

def print_tensor_name(chkpt_fname):
		reader = pywrap_tensorflow.NewCheckpointReader(chkpt_fname)
		var_to_shape_map = reader.get_variable_to_shape_map()
		print("tensor_name")
		for key in var_to_shape_map:
				print("tensor_name: ", key)
				print(reader.get_tensor(key)) # Remove this is you want to print only variable names

python frozen_graph.py 
	--input_checkpoint=tf_files/inception/ 
	--output_graph=frozen_graph.pb 
	--input_names=DecodeJpeg 
	--output_names=final_result

4,run_count_ops.sh——print all the oprations of the model and summary,you could use tensboard command