-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathcallback_utils.py
141 lines (118 loc) · 5.43 KB
/
callback_utils.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
"""
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import cv2 as cv
import numpy as np
import skimage
from PIL import Image
import tensorflow as tf
from nets.dce_net import enhance_np
data = skimage.data
#
# class tfboard_loss_callback(tf.keras.callbacks.Callback):
# """
#
# """
# def __init__(self, log_dir='./logs/loss/', losser=None, session=None):
# super(tfboard_loss_callback, self).__init__()
# self.losser = losser
# self.session = session
# self.writer = tf.summary.FileWriter(log_dir)
# self.sumary_loss1 = tf.summary.scalar('loss_1', self.losser[7])
# self.sumary_loss2 = tf.summary.scalar('loss_2', self.losser[8])
# self.sumary_loss3 = tf.summary.scalar('loss_3', self.losser[9])
# self.sumary_loss4 = tf.summary.scalar('loss_4', self.losser[10])
#
#
# def on_epoch_end(self, epoch, logs=None):
# """
#
# :param epoch:
# :param logs:
# :return:
# """
# loss1, loss2, loss3, loss4 = self.session.run([self.sumary_loss1,
# self.sumary_loss2,
# self.sumary_loss3,
# self.sumary_loss4])
# self.writer.add_summary(loss1, epoch)
# self.writer.add_summary(loss2, epoch)
# self.writer.add_summary(loss3, epoch)
# self.writer.add_summary(loss4, epoch)
def colormap_jet(img):
min = img.min()
if min < 0:
img -= min
img *= (255.0/img.max())
temp = cv.applyColorMap(np.uint8(img), 2)
out = temp # cv.cvtColor(temp, cv.COLOR_BGR2RGB)
return out
class conv2d_callback(tf.keras.callbacks.Callback):
def __init__(self, log_dir='./logs/tmp/', feed_inputs_display=None):
super(conv2d_callback, self).__init__()
self.seen = 0
self.feed_inputs_display = feed_inputs_display
self.writer = tf.summary.FileWriter(log_dir)
def custom_set_feed_input_to_display(self, feed_inputs_display):
self.feed_inputs_display = feed_inputs_display
# copied from the above answers;
def make_image(self, numpy_img):
height, width, channel = numpy_img.shape
image = Image.fromarray(numpy_img)
import io
output = io.BytesIO()
image.save(output, format='PNG')
image_string = output.getvalue()
output.close()
return tf.Summary.Image(height=height, width=width, colorspace=channel, encoded_image_string=image_string)
# A callback has access to its associated model through the class property self.model.
def on_epoch_end(self, epoch, logs=None):
logs = logs or {}
self.seen += 1
if epoch % 1 == 0: # every 200 iterations or batches, plot the costumed images using TensorBorad;
summary_str = []
for i in range(len(self.feed_inputs_display)):
feature, _, _ = self.feed_inputs_display[i]
summary_str.append(tf.Summary.Value(tag='Compare{}/lowlight_img'.format(i), image=self.make_image(
feature))) # function colormap_jet(), defined above;
lowlight_img = np.expand_dims(feature, axis=0)/255.0
output_np = tf.keras.backend.get_session().run(
self.model.output, feed_dict={self.model.input: lowlight_img}
)
enhance_img = enhance_np([lowlight_img, output_np])
enhance_img = np.squeeze(enhance_img, axis=0)
#
min = enhance_img.min()
if min < 0:
enhance_img -= min
enhance_img *= (255.0 / enhance_img.max())
enhance_img = enhance_img.astype(np.uint8)
# enhance_img = cv.cvtColor(enhance_img, cv.COLOR_BGR2RGB)
summary_str.append(tf.Summary.Value(tag='Compare{}/enhanced_img'.format(i), image=self.make_image(
enhance_img)))
#
layer_names = [layer.name for layer in self.model.layers if 'conv' in layer.name]
for name in layer_names:
disp_pred, output_np = tf.keras.backend.get_session().run(
[self.model.get_layer(name).output, self.model.output], feed_dict={self.model.input: lowlight_img}
)
disp_pred = np.squeeze(disp_pred, axis=0)
# disp_pred = np.squeeze(self.model.predict_on_batch(feature), axis = 0)
disp_size = disp_pred.shape[-1]
for j in range(disp_size):
gray_img = disp_pred[:, :, j]
gray_img = np.expand_dims(gray_img, axis=-1)
summary_str.append(
tf.Summary.Value(tag='{}/img{}/out{}'.format(name, i, j), image=self.make_image(
colormap_jet(gray_img))))
self.writer.add_summary(tf.Summary(value=summary_str), global_step=self.seen)
class save_ckpt_callback(tf.keras.callbacks.Callback):
def __init__(self, saver, session, save_path):
super(save_ckpt_callback, self).__init__()
self.saver = saver
self.session = session
self.save_path = save_path
def on_epoch_end(self, epoch, logs=None):
self.saver.save(self.session, self.save_path, global_step=epoch)