-
Notifications
You must be signed in to change notification settings - Fork 5
/
ConvDAE_pred.py
170 lines (130 loc) · 5.04 KB
/
ConvDAE_pred.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
# -*- coding: utf-8 -*-
'''
【predictor for ConvDAE】
usage：
contents to modify：model path, model dir; SAVE_FLAG;
in_imgs, gt_imgs; keep_prob, feed_dict; outputs (according to act_out_fun)
notes：
Different models' "outputs_" could be different. Sometimes it won't report errors,
but the result is not as expected.In this case, return to your model definition and confirm your "outputs_"
Author: zhihong (z_zhi_hong@163.com)
Date: 20190505
Modified: zhihong_20190809
'''
# In[]:
# import modules
import os
import numpy as np
import tensorflow as tf
from time import time
import matplotlib.pyplot as plt
import matplotlib.image as plt_img
from util import my_io
#from util import my_img_evaluation as my_evl
from tensorflow.compat.v1 import ConfigProto
from tensorflow.compat.v1 import InteractiveSession
# In[]
# environment config
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
config = tf.ConfigProto()
config.gpu_options.allow_growth=True
config.gpu_options.per_process_gpu_memory_fraction = 0.8
# In[]:
# graph reset
tf.reset_default_graph()
# In[]:
# parameters config
# setting
SAVE_FLAG = 1 # flag of saving the outputs of the network's prediction
test_batch_size = 100
pic_size = [64,64] # picture size, for sml
# pic_size = [28,28] # picture size, for N-MNIST
# path
data_path = "./dataset/single_molecule_localization/sml_test.mat" # for sml
# data_path = "./dataset/N_MNIST_pic/N_MNIST_pic_test.mat" # for N-MNIST
root_model_path = "model_data/" # model's root dir
model_dir = "bsc-ConvDAE(unsup)--02-24_10-33" # model'saving dir
model_path = root_model_path + model_dir + "/"
# model
model_name = 'my_model' #model's name
model_ind = -1 #model's index
pred_res_path = './predict_res/' + data_path.split('/')[-1][0:-4] + "-" + model_dir + "/" # dir of the prediction results
if not os.path.isdir(pred_res_path) and SAVE_FLAG:
os.makedirs(pred_res_path)
# In[]
# load model
# sess = tf.Session()
sess = InteractiveSession(config=config)
sess.run(tf.global_variables_initializer())
restorer = tf.train.import_meta_graph(model_path+model_name+'.meta')
ckpt = tf.train.get_checkpoint_state(model_path)
if ckpt:
ckpt_states = ckpt.all_model_checkpoint_paths
restorer.restore(sess, ckpt_states[model_ind])
# load Ops and variables according to old model and your need
graph = tf.get_default_graph()
inputs_ = graph.get_tensor_by_name("inputs/inputs_:0")
mask_prob = graph.get_tensor_by_name("inputs/Placeholder_1:0")
targets_ = graph.get_tensor_by_name("inputs/targets_:0")
keep_prob = graph.get_tensor_by_name("inputs/Placeholder:0") # for dropout
outputs_ = graph.get_tensor_by_name("outputs/outputs_:0") # for act_fun = relu
# outputs_ = graph.get_tensor_by_name("outputs/conv2d/Relu:0")
# outputs_ = graph.get_tensor_by_name("outputs/conv2d/Tanh:0") #for act_fun = tanh
cost = graph.get_tensor_by_name("loss/Mean:0")
# In[]:
# load data
pic_test_data = my_io.load_mat(data_path)
pic_test_x = pic_test_data['data'].astype('float32')
print('pic_test_x: ', pic_test_x.shape)
in_imgs = pic_test_x
#num_selected = 200
#test_idx = np.linspace(0,len(pic_test_x)-1,num_selected).astype('int32')
#in_imgs = pic_test_x[test_idx]
#gt_imgs = pic_test_y[test_idx]
# In[]:
# prediction
ind = 0
mean_cost = 0
time_cost = 0
reconstructed = np.zeros(in_imgs.shape, dtype='float32')
for batch_x, _ in my_io.batch_iter(test_batch_size,in_imgs, in_imgs, shuffle=False):
x = batch_x.reshape((-1, *pic_size, 1))
feed_dict = {inputs_: x, keep_prob:1.0, mask_prob:0.0} #for dropout
# feed_dict = {inputs_: x, targets_: y} #for non dropout
time1 = time()
res_imgs = sess.run(outputs_, feed_dict=feed_dict)
time2 = time()
time_cost += (time2 - time1)
res_imgs = np.squeeze(res_imgs)
reconstructed[ind*test_batch_size:(ind+1)*test_batch_size] = res_imgs
ind += 1
time_cost = time_cost/len(in_imgs)
print('\nmean time cost(ms):%f\n'%(time_cost*1e3))
# In[]:
# save the prediction results
if SAVE_FLAG:
# np.save(pred_res_path+'pred_res',reconstructed) # save pics in the format of .npy
# print('\nreconstruction data saved to : \n',pred_res_path+'pred_res.npy' )
data_save = {'reconstructed': reconstructed} # save pics in the format of .mat
my_io.save_mat(pred_res_path +'recon.mat', data_save)
for i in range(len(reconstructed)): # save pics in the format of .png
plt_img.imsave(pred_res_path+'png'+str(i)+'.png', reconstructed[i], cmap=plt.cm.gray)
print('\nreconstruction data saved to : \n',pred_res_path)
# In[]:
# illustrate the results
start = 0
end = len(reconstructed)-1
idx = np.linspace(start, end, 10).astype('int32') # show 10 results at equal intervals
in_images = in_imgs[idx]
recon_images = reconstructed[idx]
fig, axes = plt.subplots(nrows=2, ncols=10, sharex=True, sharey=True, figsize=(20,4))
for images, row in zip([in_images, recon_images], axes):
for img, ax in zip(images, row):
ax.imshow(img.reshape((*pic_size)), cmap='gray')
ax.get_xaxis().set_visible(False)
ax.get_yaxis().set_visible(False)
fig.tight_layout(pad=0.1)
plt.show()
# In[24]:
# release
sess.close()