-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathCNN_Depth1.py
214 lines (165 loc) · 8.94 KB
/
CNN_Depth1.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
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
from keras.models import Model
from keras.layers import Input, Concatenate
from keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau, Callback
import os
import numpy as np
import time
#from google.colab import files
from keras.preprocessing.image import ImageDataGenerator
import pandas as pd
import tensorflow as tf
from CustomGenerator import generate_generator_FastCoding
from CustomGenerator import MultiChannelDataGen
from LoadLFImages_Backup import LoadFiftySixViewsDepth1
from ML_Schemes import Kim_NN_CTU_FV_D0, Basic_NN_CTU_rate
from ML_Schemes import Basic_NN_CTU_FV_D0
from ML_Schemes import Basic_NN_CTU_D0
from ML_Schemes import NN_FV
from UtilityFuncs import Stats_Split_Non_Split
import random
def Depth1_CNN(loopidx,CTU_FLAG,MOTION_FLAG,CTU_NN_FLAG,rate_list,DBlist,PRINT_STATUS,PRINT_STATS,TestInfo,EPOCHS,outputFolder,MULTI_CTU_FLAG,METHODOLOGY,PreDefinedTest,DEBUG):
LF_list = [28] # Total LF images in Dataset
counter=0
for DB in DBlist:
TotLF = LF_list[counter]
counter=counter+1
for rate in rate_list:
Record_Rates = []
if (PRINT_STATUS):
print("Depth 1: Processing Rate: %s, FLAGS CTU_FLAG; %d, CTU_NN_FLAG: %d , TEST: %s" % (
rate, CTU_FLAG, CTU_NN_FLAG, TestInfo))
IMAGE_SIZE = (32, 32)
batch_size = 32
pathDrive = outputFolder +'Depth1/QP_' + str(rate) + '/'
if not os.path.exists(pathDrive):
os.makedirs(pathDrive)
#sys.path.append(pathDrive)
physical_devices = tf.config.list_physical_devices('GPU')
tf.config.experimental.set_memory_growth(physical_devices[0], True)
if (PreDefinedTest):
# Pre defined three Test Images selected based on distribution of split and non split ratio
randomTestData_22 = [['4', '10', '13'], ['14', '23', '27'], ['5', '7', '16']]
randomTestData_27 = [['5', '10', '25'], ['15', '19', '22'], ['6', '12', '26']]
randomTestData_32 = [['9', '17', '19'], ['6', '12', '22'], ['3', '6', '12']]
if TestInfo == 'A':
index=0
elif TestInfo == 'B':
index=1
elif TestInfo == 'C':
index=2
if (DEBUG):
TotLF = 4
randomTestData = ['2']
elif (rate == '22'):
randomTestData = randomTestData_22[index]
elif (rate == '27'):
randomTestData = randomTestData_27[index]
elif (rate == '32'):
randomTestData = randomTestData_32[index]
else:
# Defining Random 3 LF images as test Images
randomTestData = []
for rn in range(3):
ranvalue = random.randint(1, TotLF)
randomTestData.append(str(ranvalue))
ctuPath = 'C:\\PHD\\Dataset2022\\Depth1\\CTU_' + rate + '\\'
metaPath = 'C:\\PHD\\Dataset2022\\Depth1\\META_' + rate + '\\'
# Read 56 views out of 81 views
Record=LoadFiftySixViewsDepth1(TotLF, ctuPath, metaPath, rate, PRINT_STATUS,METHODOLOGY,DB)
Record_Rates.extend(Record)
# From Here
rate1Df = pd.DataFrame(np.array(Record_Rates),
columns=['FileName', 'LFname', 'lfR', 'lfC', 'rateValue', 'CTUno', 'CurrPU', 'CurrSL', 'N1PU',
'N2PU', 'N3PU', 'N4PU', 'N1SL', 'N2SL', 'N3SL', 'N4SL', 'N1_CTU', 'N2_CTU', 'N3_CTU',
'N4_CTU', 'viewNumber','PartNo'])
print("Dataset is read with Total Samples: ",len(rate1Df))
# Splitting test and train data
train_df = rate1Df[~rate1Df['LFname'].isin(randomTestData)]
validate_df = rate1Df[rate1Df['LFname'].isin(randomTestData)]
train_df = train_df.reset_index(drop=True) # this reset the list so that it start from 0 to its end
validate_df = validate_df.reset_index(drop=True) # this reset the list so that it start from 0 to its e
# Stats of Test and Train data
if(PRINT_STATS):
Stats_Split_Non_Split(rate1Df, "[Input Data :]")
Stats_Split_Non_Split(train_df, "[Training Data:] ")
Stats_Split_Non_Split(validate_df, "[ Test Data :]")
print("The Test Samples are: ",randomTestData)
#************************* Custom generator code ************************
train_datagen = ImageDataGenerator(
rescale=1. / 255,
)
valid_datagen = ImageDataGenerator(
rescale=1. / 255,
)
if(MULTI_CTU_FLAG):
dictionaryDF = {
'Cur_CTU': 'FileName',
'N1_CTU': 'N1_CTU',
'N2_CTU': 'N2_CTU',
'N3_CTU': 'N3_CTU',
'N4_CTU': 'N4_CTU',
'N1PU': 'N1PU',
'N2PU': 'N2PU',
'N3PU': 'N3PU',
'N4PU': 'N4PU',
'N1SL': 'N1SL',
'N2SL': 'N2SL',
'N3SL': 'N3SL',
'N4SL': 'N4SL',
}
IMAGE_SIZE_MCTU = (32, 32,1)
# N1PU N2PU N3PU N4PU N1SL N2SL N3SL N4SL
inputgenerator = MultiChannelDataGen(train_df,
X_col=dictionaryDF,
y_col={'type': 'CurrSL'},
batch_size=batch_size, input_size=IMAGE_SIZE_MCTU)
testgenerator = MultiChannelDataGen(validate_df,
X_col=dictionaryDF,
y_col={'type': 'CurrSL'},
batch_size=batch_size, input_size=IMAGE_SIZE_MCTU)
else:
inputgenerator = generate_generator_FastCoding(train_datagen, train_df,IMAGE_SIZE,CTU_FLAG,CTU_NN_FLAG,MOTION_FLAG,batch_size)
testgenerator = generate_generator_FastCoding(valid_datagen, validate_df,IMAGE_SIZE,CTU_FLAG,CTU_NN_FLAG,MOTION_FLAG,batch_size)
# ******************************* [Network Defination] **********************************
if (MOTION_FLAG):
rate_input_shape = (18,) # have to check
else:
rate_input_shape = (9,)
rate_input = Input(shape=rate_input_shape)
img_input_shape= (32,32,1)
# Selection of ML scheme for prediction of non key views
if (CTU_NN_FLAG): # CTU + Feature vector case
z,InpArray=Basic_NN_CTU_FV_D0(img_input_shape,rate_input) # Basic NN CTU + Feature vector case
elif (CTU_FLAG):# CTU only case
z,InpArray=Basic_NN_CTU_D0(img_input_shape) # Basic NN CTU only case
elif (MULTI_CTU_FLAG):# CTU only case
img_input_shape= (32,32,5) # The input CTU is 64x64
z,InpArray=Basic_NN_CTU_D0(img_input_shape) # Basic NN CTU only case
else: # only feature vector case
z,InpArray=NN_FV(rate_input) # NN with Feature vector case
model = Model(inputs=InpArray, outputs=z)
model.summary()
model.compile(optimizer='rmsprop', loss='binary_crossentropy', metrics=[
'acc']) # , metrics=[mean_iou]) # The mean_iou metrics seens to leak train and test values...
class TimeHistory(Callback):
def on_train_begin(self, logs={}):
self.times = []
def on_epoch_begin(self, epoch, logs={}):
self.epoch_time_start = time.time()
def on_epoch_end(self, epoch, logs={}):
self.times.append(time.time() - self.epoch_time_start)
FLAGS_VALUE = 'Rate_' + rate + '_TEST_' + TestInfo + '_FLAGS_' + str(CTU_FLAG) + '_' + str(CTU_NN_FLAG)
print(FLAGS_VALUE)
modelPath = pathDrive + FLAGS_VALUE + '_model_val_acc_best.h5'
checkpoint = ModelCheckpoint(filepath=modelPath,
verbose=1, monitor='val_acc',
save_best_only=True, mode='max')
history = model.fit(inputgenerator, steps_per_epoch=np.floor(len(train_df) / batch_size), epochs=EPOCHS, shuffle=True,
callbacks=[checkpoint], verbose=1, validation_data=testgenerator,
validation_steps=np.floor(len(validate_df) / batch_size))
hist_df = pd.DataFrame(history.history)
hist_csv_file = pathDrive + FLAGS_VALUE + 'Summary' + '.csv'
with open(hist_csv_file, mode='w') as t:
hist_df.to_csv(t)
# To HERE
print(" ******* Depth 1 Code is done ********")