-
Notifications
You must be signed in to change notification settings - Fork 0
/
common.py
264 lines (201 loc) · 7.26 KB
/
common.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
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
import csv
import numpy as np
import os
import tensorflow as tf
import matplotlib.pyplot as plt
import sys
import shutil
import itertools
from sklearn import metrics
LOCAL_MODE = True
if LOCAL_MODE:
TRAIN_PATH = sys.argv[1]
MODELS_FOLDER = sys.argv[2]
assert os.path.isdir(TRAIN_PATH), "arg 1 must be a folder!"
os.makedirs(MODELS_FOLDER, exist_ok=True)
assert os.path.isdir(MODELS_FOLDER), "arg 2 must be a folder!"
# !mkdir -p /content/drive/Shareddrives/deep_learning/models
"""# ADDESTRAMENTO
## No augmentation
"""
BATCH_SIZE = 64
WIDTH, HEIGHT = 128, 65
EPOCHS = 50
# helper class to switch between color-modes
class Colors:
class ColorMode:
def __init__(self, keyword: str, channels: int):
self.keyword = keyword
self.channels = channels
# Define color modes as class instances
RGB = ColorMode('rgb', 3)
GRAYSCALE = ColorMode('grayscale', 1)
COLOR_MODE = Colors.GRAYSCALE
# compile_model compiles a model
def compile_model(model):
import tensorflow_addons as tfa
f1 = tfa.metrics.F1Score(num_classes=N_CLASSES)
model.compile(
optimizer='adam',
loss='categorical_crossentropy',
metrics=[
f1,
"accuracy"
],
)
def get_model_folder(name: str):
return os.path.join(MODELS_FOLDER, name)
def get_model_weights_path(name: str):
return os.path.join(get_model_folder(name), "model.keras")
def plot(model: tf.keras.Model, history: any, X, Y, d: tf.data.Dataset, name: str):
out_folder = get_model_folder(name)
if history is not None:
# Plot training history f1_score
plt.figure(figsize=(3.66, 3.66))
plt.plot(np.mean(history.history['f1_score'], axis=1), label='avg f1_score')
plt.plot(np.mean(history.history['val_f1_score'], axis=1), label='avg val_f1_score')
plt.xlabel('Epoch')
plt.ylabel('F1-score')
plt.ylim([0, 1])
plt.legend(loc='lower right')
plt.show()
plt.draw()
plt.savefig(os.path.join(out_folder, "learning_history-f1_score.png"), dpi=96*5)
plt.close()
# Plot training history loss
plt.figure(figsize=(3.66, 3.66))
plt.plot(history.history['loss'], label='loss')
plt.plot(history.history['val_loss'], label='val_loss')
plt.xlabel('Epoch')
plt.ylabel('Loss')
plt.legend(loc='lower right')
plt.show()
plt.draw()
plt.savefig(os.path.join(out_folder, "learning_history-loss.png"), dpi=96*5)
plt.close()
# Plot training history accuracy
plt.figure(figsize=(3.66, 3.66))
plt.plot(history.history['accuracy'], label='accuracy')
plt.plot(history.history['val_accuracy'], label='val_accuracy')
plt.xlabel('Epoch')
plt.ylabel('Accuracy')
plt.ylim([0, 1])
plt.legend(loc='lower right')
plt.show()
plt.draw()
plt.savefig(os.path.join(out_folder, "learning_history-acc.png"), dpi=96*5)
plt.close()
# PLot confusion matrix
preds = model.predict(X)
Y_pred = np.argmax(preds, axis=1)
cm = metrics.confusion_matrix(Y, Y_pred, normalize='true')
#cm = np.trunc(cm * 10 ** 2) / (10 ** 2)
# LOG this should be equal to the original one
# correct_predictions = sum(1 for p, t in zip(Y_pred, Y_test) if p == t)
# total_predictions = len(Y_pred)
# accuracy = correct_predictions / total_predictions
# print("Accuracy: ", accuracy)
# LOG
plt.figure(figsize=(10, 10))
plt.imshow(cm, interpolation='nearest', cmap=plt.cm.Blues)
plt.title('Confusion Matrix')
plt.colorbar()
tick_marks = np.arange(N_CLASSES)
plt.xticks(tick_marks, CLASSES, rotation=45)
plt.yticks(tick_marks, CLASSES)
thresh = cm.max() / 2.
for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):
plt.text(j, i, "{:.2f}".format(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()
plt.draw()
plt.savefig(os.path.join(out_folder, "confusion_matrix.png"), dpi=96*8)
plt.close()
# Save metrics on eval test on file
res = model.evaluate(d, return_dict=True)
with open(os.path.join(out_folder, "metrics.csv"), 'w') as csvfile:
writer = csv.DictWriter(csvfile, fieldnames=list(res.keys()))
writer.writeheader()
writer.writerows([res])
# F1-scores
with open(os.path.join(out_folder, "f1.csv"), 'w') as file:
file.write("class,f1\n")
for i, c in enumerate(CLASSES):
file.write(c + "," + "{:.3f}".format(res["f1_score"][i]) + "\n")
# train trains a model and put its weight in the specified output path
def train(model: any, name: str):
# define useful callbacks
early_stop = tf.keras.callbacks.EarlyStopping(
monitor='loss',
min_delta=0.02,
patience=6,
)
save_best_model = tf.keras.callbacks.ModelCheckpoint(
filepath=get_model_weights_path(name),
monitor='val_loss',
save_best_only=True,
save_weights_only=True,
)
csv_logger = tf.keras.callbacks.CSVLogger(
os.path.join(get_model_folder(name), "model.stats.csv"),
append=True
)
# Train the model
history = model.fit(
train_dataset,
epochs=EPOCHS,
validation_data=val_dataset,
validation_steps=len(val_dataset),
callbacks=[
csv_logger,
early_stop,
save_best_model
],
verbose=1,
workers=4
)
return history
def evalutate(model: tf.keras.Model, name: str):
model_out_folder = get_model_folder(name)
if os.path.exists(model_out_folder):
shutil.rmtree(model_out_folder)
os.makedirs(model_out_folder, exist_ok=True)
history = train(model, name)
plot(model, history, X_val, Y_val, val_dataset, name)
"""La roba vera per l'addestrament insomma"""
# Load the dataset without validation splitting
dataset = tf.keras.utils.image_dataset_from_directory(
TRAIN_PATH,
image_size=(HEIGHT, WIDTH),
color_mode=COLOR_MODE.keyword,
batch_size=BATCH_SIZE,
label_mode="categorical",
shuffle=True,
seed=0xcafebabe
)
"""devide the dataset and log info"""
CLASSES = dataset.class_names
N_CLASSES = len(dataset.class_names)
# Calculate the number of validation samples
N_SAMPLES = dataset.cardinality().numpy()
VALIDATION_TEST_SAMPLES = int(0.35 * N_SAMPLES) # 35% of data for validation and test
VALIDATION_SAMPLES = int(0.5 * VALIDATION_TEST_SAMPLES)
# Split the dataset into training and validation
val_test_dataset = dataset.take(VALIDATION_TEST_SAMPLES)
train_dataset = dataset.skip(VALIDATION_TEST_SAMPLES)
test_dataset = val_test_dataset.take(VALIDATION_SAMPLES)
val_dataset = val_test_dataset.skip(VALIDATION_SAMPLES)
def generate_eval_matrixes(dataset: tf.data.Dataset):
X = []
Y = []
for images, labels in dataset:
for image in images:
X.append(np.array(image.numpy().tolist())) # append list
for label in labels:
Y.append(np.argmax(label.numpy(), axis=0))
X = tf.constant(X)
return X, Y
X_val, Y_val = generate_eval_matrixes(val_dataset)
X_test, Y_test = generate_eval_matrixes(test_dataset)