-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathskin_cancer
287 lines (244 loc) · 11.5 KB
/
skin_cancer
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
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
# -*- coding: utf-8 -*-
"""
Created on Mon Sep 11 10:13:59 2023
@author: LI
"""
# import system libs
import os
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import itertools
# import Deep learning Libraries
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense
from tensorflow.keras.models import Model
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from sklearn.metrics import confusion_matrix, classification_report
from tensorflow.keras.optimizers import Adamax
from tensorflow.keras.callbacks import EarlyStopping, ReduceLROnPlateau
from sklearn.model_selection import train_test_split
from keras.utils.np_utils import to_categorical
from imblearn.over_sampling import RandomOverSampler
import numpy as np
import pandas as pd
from matplotlib import rcParams
model_path = 'D:/tf/FPGA/tensorflow/skin/conv_test/cov_test.h5'
data_dir = 'D:/tf/FPGA/tensorflow/skin/hmnist_28_28_RGB.csv'
data = pd.read_csv(data_dir)
data.head()
Label = data["label"]
Data = data.drop(columns=["label"])
data["label"].value_counts()
label_counts = data["label"].value_counts()
print(label_counts)
X_total = Data
y_total = Label
oversample = RandomOverSampler()
Data, Label = oversample.fit_resample(Data, Label)
Data = np.array(Data).reshape(-1, 28, 28, 3)
print('Shape of Data :', Data.shape)
Label = np.array(Label)
Label
#Convert abbreviations to it's words
classes = {0: ('akiec', 'Actinic keratoses and intraepithelial carcinomae'),
1:('bcc' , ' basal cell carcinoma'),
2 :('bkl', 'benign keratosis-like lesions'),
3: ('df', 'dermatofibroma'),
4: ('nv', ' melanocytic nevi'),
5: ('vasc', ' pyogenic granulomas and hemorrhage'),
6: ('mel', 'melanoma')}
# train_test_split
X_train , X_test , y_train , y_test = train_test_split(Data , Label , test_size = 0.2 , random_state = 49)
#Create Image Data Generation
datagen = ImageDataGenerator(rescale=(1./255)
,rotation_range=10
,zoom_range = 0.1
,width_shift_range=0.1
,height_shift_range=0.1)
testgen = ImageDataGenerator(rescale=(1./255))
#Create ReduceLROnPlateau to learning rate reduction
learning_rate_reduction = ReduceLROnPlateau(monitor='val_accuracy'
, patience = 2
, verbose=1
,factor=0.5
, min_lr=0.00001)
'''
X_train,X_val,y_train,y_val =train_test_split(X_train, y_train, test_size=0.25, random_state=42)
test_data= test_generator.flow_from_dataframe(dataframe=test,x_col="image_full_name",y_col="dx",
directory="../HAM1000_images",
shuffle=False,batch_size=1,class_mode=None,target_size=(28,28))
val_data=test_generator.flow_from_dataframe(dataframe=val,x_col="image_full_name",y_col="dx",
directory="../HAM1000_images",
batch_size=10,shuffle=False,class_mode="categorical",target_size=(28,28))
train_data= train_generator.flow_from_dataframe(dataframe=train,x_col="image_full_name",y_col="dx",
batch_size=10,directory="../HAM1000_images",
shuffle=True,class_mode="categorical",target_size=(28,28))
'''
def cnn():
use_bias = False
#Model Structure
model = keras.models.Sequential()
model.add(keras.layers.Input(shape=[28, 28, 3]))
model.add(keras.layers.Conv2D(4, (3, 3), activation='relu', padding='same',name='cov1',use_bias = use_bias))
model.add(keras.layers.MaxPooling2D(pool_size=(2,2),name='pool1'))
model.add(keras.layers.Conv2D(8, (3, 3), activation='relu', padding='same',name='cov2',use_bias=use_bias))
model.add(keras.layers.MaxPooling2D(pool_size=(2,2),name='pool2'))
model.add(keras.layers.Conv2D(16, (3, 3), activation='relu', padding='same',name='cov3',use_bias = use_bias))
model.add(keras.layers.Flatten(name='Flatten'))
model.add(keras.layers.Dense(units=32, activation='relu', name='dense',use_bias = use_bias))
model.add(keras.layers.Dense(units=7, activation='softmax', name='softmax',use_bias = use_bias))
return model
# Set a learning rate annealer
if os.path.exists(model_path):
model = tf.keras.models.load_model(filepath=model_path)
print(model.summary())
else:
model.compile(Adamax(learning_rate= 0.001), loss= 'categorical_crossentropy', metrics= ['accuracy'])
early_stopping = EarlyStopping(monitor='val_loss', patience=5, restore_best_weights=True)
history = model.fit(X_train ,
y_train,
epochs=100,
batch_size=10,
validation_data=(X_test , y_test) ,
callbacks=[learning_rate_reduction,early_stopping])
model.save("D:/tf/FPGA/tensorflow/skin/conv_test/conv_test/10_7.h5")
print(model.summary())
train_class_counts = [np.sum(np.argmax(y_train, axis=1) == i) for i in range(len(classes))]
test_class_counts = [np.sum(np.argmax(y_test, axis=1) == i) for i in range(len(classes))]
class_names = [classes[i][0] for i in range(len(classes))]
y_true = np.array(y_test)
y_pred = model.predict(X_test)
y_pred = np.argmax(y_pred , axis=1)
y_true = np.argmax(y_true , axis=1)
classes_labels = []
for key in classes.keys():
classes_labels.append(key)
print(classes_labels)
cm = confusion_matrix(y_true, y_pred, labels=classes_labels)
#cm = cm = confusion_matrix(y_true, y_pred, labels=classes_labels)
proportion=[]
for i in cm:
for j in i:
temp=j/(np.sum(i))
proportion.append(temp)
# print(np.sum(confusion_matrix[0]))
#print(proportion)
pshow=[]
for i in proportion:
pt="%.2f%%" % (i * 100)
pshow.append(pt)
proportion=np.array(proportion).reshape(7,7) # reshape
pshow=np.array(pshow).reshape(7,7)
#print(pshow)
config = {
"font.family":'Times New Roman',
}
rcParams.update(config)
plt.imshow(proportion, interpolation='nearest', cmap=plt.cm.PuRd)
# (：'Greys', 'Purples', 'Blues', 'Greens', 'Oranges', 'Reds','YlOrBr', 'YlOrRd',
# 'OrRd', 'PuRd', 'RdPu', 'BuPu','GnBu', 'PuBu', 'YlGnBu', 'PuBuGn', 'BuGn', 'YlGn')
plt.title('confusion_matrix')
plt.colorbar()
tick_marks = np.arange(len(classes))
plt.xticks(tick_marks, classes,fontsize=12)
plt.yticks(tick_marks, classes,fontsize=12)
thresh = cm.max() / 2.
iters = np.reshape([[[i,j] for j in range(7)] for i in range(7)],(cm.size,2))
for i, j in iters:
if(i==j):
plt.text(j, i - 0.12, format(cm[i, j]), va='center', ha='center', fontsize=12,color='white',weight=5)
plt.text(j, i + 0.12, pshow[i, j], va='center', ha='center', fontsize=12,color='white')
else:
plt.text(j, i-0.12, format(cm[i, j]),va='center',ha='center',fontsize=12)
plt.text(j, i+0.12, pshow[i, j], va='center', ha='center', fontsize=12)
plt.ylabel('True label',fontsize=16)
plt.xlabel('Predict label',fontsize=16)
plt.tight_layout()
plt.show()
for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):
plt.text(j, i, cm[i, j], horizontalalignment= 'center', color= 'white' if cm[i, j] > thresh else 'black')
plt.tight_layout()
plt.ylabel('True Label')
plt.xlabel('Predicted Label')
plt.show()
all_layers = model.layers
dense_layer_count = 0
for layer in all_layers:
if isinstance(layer, tf.keras.layers.Dense):
weights = layer.get_weights()
if len(weights) > 0:
dense_layer_count += 1
layer_name = f'dense_layer_{dense_layer_count}'
filename = f'{layer_name}_weights.txt'
np.savetxt(filename, weights[0], delimiter=' ', fmt='%.16f')
print(f"Saved weights of {layer_name} to {filename}")
input_image = X_test[0]
for channel in range(3):
channel_data = input_image[:, :, channel]
filename = f'input_image_channel_{channel}.txt'
np.savetxt(filename, channel_data, delimiter='\n', fmt='%f')
middle = Model(inputs=model.input, outputs=model.get_layer('cov1').output)
results = middle.predict(X_test)[0]
for i in range(results.shape[-1]):
result = results[:, :, i]
filename = f'D:/tf/FPGA/tensorflow/skin/conv_test/conv/conv1_kernel_{i + 1}.csv'
np.savetxt(filename, result, delimiter=',')
middle = Model(inputs=model.input, outputs=model.get_layer('cov1').output)
result = middle.predict(X_test)[0]
result = result.reshape(result.shape[0], -1)
np.savetxt('D:/tf/FPGA/tensorflow/skin/conv_test/test/conv1.csv', result, delimiter=',')
middle = Model(inputs=model.input, outputs=model.get_layer('pool1').output)
result = middle.predict(X_test)[0]
result = result.reshape(result.shape[0], -1)
np.savetxt('D:/tf/FPGA/tensorflow/skin/conv_test/test/pool1.csv', result, delimiter=',')
middle = Model(inputs=model.input, outputs=model.get_layer('cov2').output)
result = middle.predict(X_test)[0]
result = result.reshape(result.shape[0], -1)
np.savetxt('D:/tf/FPGA/tensorflow/skin/conv_test/test/cov2.csv', result, delimiter=',')
middle = Model(inputs=model.input, outputs=model.get_layer('pool2').output)
result = middle.predict(X_test)[0]
result = result.reshape(result.shape[0], -1)
np.savetxt('D:/tf/FPGA/tensorflow/skin/conv_test/test/pool2.csv', result, delimiter=',')
middle = Model(inputs=model.input, outputs=model.get_layer('cov3').output)
result = middle.predict(X_test)[0]
result = result.reshape(result.shape[0], -1)
np.savetxt('D:/tf/FPGA/tensorflow/skin/conv_test/conv3.csv', result, delimiter=',')
middle = Model(inputs=model.input, outputs=model.get_layer('Flatten').output)
result = middle.predict(X_test)[0]
result = result.reshape(result.shape[0], -1)
np.savetxt('D:/tf/FPGA/tensorflow/skin/conv_test/Flatten.csv', result, delimiter=',')
middle = Model(inputs=model.input, outputs=model.get_layer('dense').output)
result = middle.predict(X_test)[0]
result = result.reshape(result.shape[0], -1)
np.savetxt('D:/tf/FPGA/tensorflow/skin/conv_test/dense.csv', result, delimiter=',')
middle = Model(inputs=model.input, outputs=model.get_layer('softmax').output)
result = middle.predict(X_test)[0]
result = result.reshape(result.shape[0], -1)
np.savetxt('D:/tf/FPGA/tensorflow/skin/conv_test/softmax.csv', result,delimiter=',')
conv_weights = []
#dense_weights = []
for layer in model.layers:
if isinstance(layer, keras.layers.Conv2D):
weights = layer.get_weights()
if weights:
conv_weights.append(weights)
for i, layer_weights in enumerate(conv_weights):
for j, kernel_weights in enumerate(layer_weights):
kernel_shape = kernel_weights.shape
for k in range(kernel_shape[-1]):
layer_name = f'conv{i + 1}_kernel{j + 1}_filter{k + 1}'
filename = f'{layer_name}_weights.txt'
filter_weights = kernel_weights[:, :, :, k]
flattened_weights = filter_weights.flatten()
np.savetxt(filename, flattened_weights, delimiter=' ', fmt='%.16f')
for i, weights in enumerate(conv_weights):
layer_name = f'conv{i + 1}'
filename = f'{layer_name}_weights.txt'
flattened_weights = weights[0].flatten()
np.savetxt(filename, flattened_weights, delimiter=' ', fmt='%.16f')
for i, weights in enumerate(dense_weights):
layer_name = f'dense{i + 1}'
filename = f'{layer_name}_weights.txt'
np.savetxt(filename, weights[0], delimiter=' ', fmt='%.16f')