-
Notifications
You must be signed in to change notification settings - Fork 0
/
Pre_Trained_Pipeline.py
174 lines (162 loc) · 6.97 KB
/
Pre_Trained_Pipeline.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
import numpy as np
import tensorflow as tf
from tensorflow import keras
import os
import cv2
import matplotlib.pyplot as plt
def Pre_Process(prompt,x=None,ct=None):
clahe = cv2.createCLAHE(clipLimit=4.0, tileGridSize=(8,8))
ker = np.ones((2,2))
if prompt==1:
clahe.apply(x)
x = cv2.resize(x,(224,224))
x = cv2.cvtColor(x,cv2.COLOR_GRAY2RGB)
return x.reshape((1,224,224,3)),None
elif prompt==2:
clahe.apply(ct)
ct = cv2.morphologyEx(ct, cv2.MORPH_OPEN, ker)
ct = cv2.resize(ct,(224,224))
ct = cv2.cvtColor(ct,cv2.COLOR_GRAY2RGB)
return None,ct.reshape((1,224,224,3))
else:
clahe.apply(x)
clahe.apply(ct)
ct = cv2.morphologyEx(ct, cv2.MORPH_OPEN, ker)
x = cv2.resize(x,(224,224))
ct = cv2.resize(ct,(224,224))
x = cv2.cvtColor(x,cv2.COLOR_GRAY2RGB)
ct = cv2.cvtColor(ct,cv2.COLOR_GRAY2RGB)
return x.reshape((1,224,224,3)),ct.reshape((1,224,224,3))
def Task_Specific_Embed(Save_Folder_Directory,x=None,ct=None):
if x is not None and ct is None:
x_model = keras.models.load_model(Save_Folder_Directory+'/X_Embed_Task_EfficientNetB0.h5',compile=False)
task_x = keras.models.load_mode(Save_Folder_Directory+'X_Ray_Task.h5',compile=False)
x_layers = x_model.layers[:-1]
task_x_layers = task_x.layers[2:-1]
for i in x_layers:
x = i(x)
for k in task_x_layers:
x = k(x)
return x,None
elif x is None and ct is not None:
ct_model = keras.models.load_model(Save_Folder_Directory+'/CT_Embed_Task_EfficientNetB1.h5',compile=False)
task_ct = keras.models.load_model(Save_Folder_Directory+'/CT_Scan_Task.h5',compile=False)
ct_layers = ct_model.layers[:-1]
task_ct_layers = task_ct.layers[1:-1]
for j in ct_layers:
ct = j(ct)
for l in task_ct_layers:
ct = l(ct)
return None,ct
else:
x_model = keras.models.load_model(Save_Folder_Directory+'/X_Embed_Task_EfficientNetB0.h5',compile=False)
ct_model = keras.models.load_model(Save_Folder_Directory+'/CT_Embed_Task_EfficientNetB1.h5',compile=False)
task_x = keras.models.load_mode(Save_Folder_Directory+'X_Ray_Task.h5',compile=False)
task_ct = keras.models.load_model(Save_Folder_Directory+'/CT_Scan_Task.h5',compile=False)
x_layers = x_model.layers[:-1]
ct_layers = ct_model.layers[:-1]
task_x_layers = task_x.layers[2:-1]
task_ct_layers = task_ct.layers[1:-1]
for i in x_layers:
x = i(x)
for j in ct_layers:
ct = j(ct)
for k in task_x_layers:
x = k(x)
for l in task_ct_layers:
ct = l(ct)
return x,ct
def Shared_Feature_Embed(Save_Folder_Directory,x=None,ct=None):
if x is not None and ct is None:
x_model = keras.models.load_model(Save_Folder_Directory+'/X_Embed_Shared_ResNet50.h5',compile=False)
shared_model = keras.models.load_model(Save_Folder_Directory+'/Shared_Model.h5',compile=False)
x_layers = x_model.layers[:-1]
sh_layers = shared_model.layers[:-1]
for i in x_layers:
x = i(x)
for k in sh_layers:
x = k(x)
return x,None
elif x is None and ct is not None:
ct_model = keras.models.load_model(Save_Folder_Directory+'/CT_Embed_Shared_ResNet50V2.h5',compile=False)
shared_model = keras.models.load_model(Save_Folder_Directory+'/Shared_Model.h5',compile=False)
ct_layers = ct_model.layers[:-1]
sh_layers = shared_model.layers[:-1]
for j in ct_layers:
ct = j(ct)
for k in sh_layers:
ct = k(ct)
return None,ct
else:
x_model = keras.models.load_model(Save_Folder_Directory+'/X_Embed_Shared_ResNet50.h5',compile=False)
ct_model = keras.models.load_model(Save_Folder_Directory+'/CT_Embed_Shared_ResNet50V2.h5',compile=False)
shared_model = keras.models.load_model(Save_Folder_Directory+'/Shared_Model.h5',compile=False)
x_layers = x_model.layers[:-1]
ct_layers = ct_model.layers[:-1]
sh_layers = shared_model.layers[:-1]
for i in x_layers:
x = i(x)
for j in ct_layers:
ct = j(ct)
for k in sh_layers:
x = k(x)
ct = k(ct)
return x,ct
def Classification(Save_Folder_Directory,x=None,ct=None):
if x is None and ct is not None:
classifier_ct = keras.models.load_model(Save_Folder_Directory+'/Classifier_Head_CT.h5',compile=False)
_,ta_ct = Task_Specific_Embed(Save_Folder_Directory,x,ct)
_,sh_ct = Shared_Feature_Embed(Save_Folder_Directory,x,ct)
Feature_ct = np.concatenate((ta_ct,sh_ct),axis=1)
result_ct = classifier_ct.predict(Feature_ct)
print("Probability :-", result_ct)
if result_ct>=0.5:
print("Prediction :- COVID")
else:
print("Prediction :- Non-COVID")
elif x is not None and ct is None:
classifier_x = keras.models.load_model(Save_Folder_Directory+'/Classifier_Head_X.h5',compile=False)
ta_x,_ = Task_Specific_Embed(Save_Folder_Directory,x,ct)
sh_x,_ = Shared_Feature_Embed(Save_Folder_Directory,x,ct)
Feature_x = np.concatenate((ta_x,sh_x),axis=1)
result_x = classifier_x.predict(Feature_x)
print("Probability :-", result_x)
if result_x>=0.5:
print("Prediction :- COVID")
else:
print("Prediction :- Non-COVID")
else:
classifier_x = keras.models.load_model(Save_Folder_Directory+'/Classifier_Head_X.h5',compile=False)
classifier_ct = keras.models.load_model(Save_Folder_Directory+'/Classifier_Head_CT.h5',compile=False)
ta_x,ta_ct = Task_Specific_Embed(Save_Folder_Directory,x,ct)
sh_x,sh_ct = Shared_Feature_Embed(Save_Folder_Directory,x,ct)
Feature_x = np.concatenate((ta_x,sh_x),axis=1)
Feature_ct = np.concatenate((ta_ct,sh_ct),axis=1)
result_x = classifier_x.predict(Feature_x)
result_ct = classifier_ct.predict(Feature_ct)
for result in [result_x,result_ct]:
print("Probability :-", result)
if result>=0.5:
print("Prediction :- COVID")
else:
print("Prediction :- Non-COVID")
print(" ")
def Test_Run(Save_Folder_Directory,prompt,X_Ray_Grey_Image=None,CT_Scan_Grey_Image=None):
X_Ray_Image,CT_Scan_Image = Pre_Process(prompt,X_Ray_Grey_Image,CT_Scan_Grey_Image)
Classification(Save_Folder_Directory,X_Ray_Image,CT_Scan_Image)
user_prompt = int(input("Enter the choice of operation --> 1 for Only X-Ray, 2 for only CT-Scan or 3 for Both"))
if user_prompt==1:
x_ray = cv2.imread(input("Enter the File Path for Chest X-Ray Input Image --> "),0)
root_directory = input("Enter the Root Folder Directory Path for Each and Every saved Components of the Pre-Trained Pipeline --> ")
Test_Run(root_directory,user_prompt,x_ray,None)
elif user_prompt==2:
ct_scan = cv2.imread(input("Enter the File Path for CT-Scan Input Image --> "),0)
root_directory = input("Enter the Root Folder Directory Path for Each and Every saved Components of the Pre-Trained Pipeline --> ")
Test_Run(root_directory,user_prompt,None,ct_scan)
elif user_prompt==3:
x_ray = cv2.imread(input("Enter the File Path for Chest X-Ray Input Image --> "),0)
ct_scan = cv2.imread(input("Enter the File Path for CT-Scan Input Image --> "),0)
root_directory = input("Enter the Root Folder Directory Path for Each and Every saved Components of the Pre-Trained Pipeline --> ")
Test_Run(root_directory,user_prompt,x_ray,ct_scan)
else:
print("Wrong Input Given")