generate_inference_model

#!/usr/bin/python

import argparse
import os

import tensorflow as tf
from tensorflow.python.framework import graph_util
from tensorflow.python.framework import graph_io
from keras import backend as K

import model

parser = argparse.ArgumentParser()
parser.add_argument('--input_width', type=int, default=640, help='the input image width')
parser.add_argument('--input_height', type=int, default=360, help='the input image height')
parser.add_argument('--weights', type=str, default='weights_1.h5', help='output path for the weights file')
opt = parser.parse_args()

print(opt)

K.set_learning_phase(0)
tiramisu = model.build(opt.input_width, opt.input_height, 66, weights_path=opt.weights)
sess = K.get_session()

for n in tf.get_default_graph().as_graph_def().node:
    print(n.name)

inference_graph = graph_util.convert_variables_to_constants(sess, sess.graph.as_graph_def(), ['output'])
inference_graph = graph_util.remove_training_nodes(inference_graph) # Just in case

graph_io.write_graph(constant_graph, './', 'tiramisu.pb', as_text=False)

print('Saved the constant graph (ready for inference) at: ./tiramisu.pb')