In [1]:

    

import numpy as np
import tensorflow as tf
from crnn_model import crnn_model
import json


    

In [2]:

    

with open('crnn_model/char_dict.json') as fopen:
    char_dict = json.load(fopen)
    
with open('crnn_model/ord_map.json') as fopen:
    order_dict = json.load(fopen)


    

In [3]:

    

class Model:
    def __init__(self):
        self.X = tf.placeholder(tf.float32,(None,None,3))
        image = tf.expand_dims(self.X, 0)
        image = tf.image.resize_images(image, [32, 100])
        num_classes = 37
        net = crnn_model.ShadowNet(phase = 'Test', 
                                   hidden_nums = 256,
                                   layers_nums = 2,
                                   num_classes = num_classes)
        with tf.variable_scope('shadow'):
            net_out = net.build_shadownet(inputdata=image)
        self.decode, _ = tf.nn.ctc_beam_search_decoder(net_out, 25 * np.ones(1), merge_repeated=False)


    

In [4]:

    

tf.reset_default_graph()
sess = tf.InteractiveSession()
model = Model()
sess.run(tf.global_variables_initializer())
saver = tf.train.Saver(tf.global_variables())
saver.restore(sess, 'rename-checkpoint2/model.ckpt')


    

    




INFO:tensorflow:Restoring parameters from rename-checkpoint2/model.ckpt

In [5]:

    

import cv2
image = cv2.imread('back-car.jpeg', cv2.IMREAD_COLOR)


    

In [6]:

    

image.shape


    

    
Out[6]:





(74, 302, 3)

In [7]:

    

output = sess.run(model.decode,feed_dict={model.X:image})[0]


    

In [8]:

    

output


    

    
Out[8]:





SparseTensorValue(indices=array([[0, 0],
       [0, 1],
       [0, 2],
       [0, 3],
       [0, 4],
       [0, 5],
       [0, 6],
       [0, 7]]), values=array([15, 18, 14, 15, 18, 17, 17, 13]), dense_shape=array([1, 8]))

In [9]:

    

def sparse_tensor_to_str(sparse_tensor):
    indices = sparse_tensor.indices
    values = sparse_tensor.values
    values = np.array([order_dict[str(tmp)] for tmp in values])
    dense_shape = sparse_tensor.dense_shape

    number_lists = np.ones(dense_shape, dtype=values.dtype)
    str_lists = []
    res = []
    for i, index in enumerate(indices):
        number_lists[index[0], index[1]] = values[i]
    for number_list in number_lists:
        str_lists.append([char_dict[val] for val in number_list])
    for str_list in str_lists:
        res.append(''.join(c for c in str_list if c != '\x00'))
    return res


    

In [10]:

    

sparse_tensor_to_str(output)[0]


    

    
Out[10]:





'psopsrrn'

In [11]:

    

char_list = '0123456789abcdefghijklmnopqrstuvwxyz '


    

In [12]:

    

def sparse_tensor_to_str2(spares_tensor):

    indices = spares_tensor.indices
    values = spares_tensor.values
    dense_shape = spares_tensor.dense_shape

    number_lists = np.ones(dense_shape, dtype=values.dtype)
    str_lists = []
    res = []
    for i, index in enumerate(indices):
        number_lists[index[0], index[1]] = values[i]
    for number_list in number_lists:
        str_lists.append([char_list[val] for val in number_list])
    for str_list in str_lists:
        res.append(''.join(c for c in str_list if c != '1'))
    return res


    

In [13]:

    

sparse_tensor_to_str2(output)


    

    
Out[13]:





['fiefihhd']

In [ ]: