-
Notifications
You must be signed in to change notification settings - Fork 0
/
train.py
409 lines (333 loc) · 15.1 KB
/
train.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
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
import cv2
import os,os.path
import pandas as pd
import numpy as np
import torch
from torch.utils.data import Dataset
import argparse
import torch.nn as nn
from torch.nn import ConvTranspose2d
from torch.nn import Conv2d
from torch.nn import MaxPool2d
from torch.nn import Module
from torch.nn import ModuleList
from torch.nn import ReLU
from torchvision.transforms import CenterCrop
from torch.nn import functional as F
import torchvision
import imutils
from torch.nn import BCEWithLogitsLoss,CrossEntropyLoss
from torch.optim import Adam
from torch.utils.data import DataLoader
from sklearn.model_selection import train_test_split
from torchvision import transforms
from imutils import paths
from tqdm import tqdm
import matplotlib.pyplot as plt
import time
from pathlib import Path
from PIL import Image
import PIL
import itertools
mask_dataset="masks"
images_dataset="images"
count=0
c = 4
r=len([name for name in os.listdir(mask_dataset) if os.path.isfile(os.path.join(mask_dataset, name))])
path_list=[]
arr = np.ones((r,c)) # bboxlar için kullanılacak olan arrayın tanımlanması
flag=1
x_file=os.listdir(images_dataset)[0]
x_file=mask_dataset+"/"+x_file
img_size = cv2.imread(x_file)
a,b,_=img_size.shape
original_image_size = [a,b]
#argümanların ataması
def args_init(args):
BATCH_SIZE = args.batch_size
NUM_EPOCHS = args.num_epochs
TEST_SPLIT = args.test_split
INIT_LR = args.lr
BASE_OUTPUT = args.BASE_OUTPUT
INPUT_IMAGE_WIDTH = args.INPUT_IMAGE_WIDTH
INPUT_IMAGE_HEIGHT = args.INPUT_IMAGE_HEIGHT
TEST_PATHS = os.path.sep.join([BASE_OUTPUT, "test_paths.txt"])
MODEL_PATH_BEST = os.path.join(BASE_OUTPUT, str(INPUT_IMAGE_HEIGHT) +"_bestmodel.pth")
MODEL_PATH_LAST = os.path.join(BASE_OUTPUT, str(INPUT_IMAGE_HEIGHT) +"_lastmodel.pth")
PLOT_PATH = os.path.sep.join([BASE_OUTPUT, "plot.png"])
print("BATCH_SIZE: {}, NUM_EPOCHS: {}, LR: {}".format(BATCH_SIZE, NUM_EPOCHS, INIT_LR))
return BATCH_SIZE,NUM_EPOCHS,TEST_SPLIT,INIT_LR,INPUT_IMAGE_WIDTH,INPUT_IMAGE_HEIGHT,TEST_PATHS,MODEL_PATH_BEST,MODEL_PATH_LAST,PLOT_PATH
## 2.BÖLÜM
class CustomDataset(Dataset):
def __init__(self,imagePaths,maskPaths,transforms,bbox):
#mask ve image'leri depolar + transform
self.imagePaths=imagePaths
self.maskPaths=maskPaths
self.transforms=transforms
self.bbox=bbox
def __len__(self):
return len(self.imagePaths)
def __getitem__(self,idx):
imagePath = self.imagePaths[idx]
image = cv2.imread(imagePath)
image = cv2.cvtColor(image, cv2.COLOR_BGRA2BGR)
bbox=self.bbox[idx]
bbox=torch.Tensor(([(bbox[0]/original_image_size[1]),(bbox[1]/original_image_size[0]),(bbox[2]/original_image_size[1]),(bbox[3]/original_image_size[0])])) # 0, 1
mask = cv2.imread(self.maskPaths[idx], 0)
if self.transforms is not None:
image = self.transforms(image)
mask = self.transforms(mask)
return (image, mask, bbox)
## 3.BÖLÜM
#U-Net mimarisi + Conv Residual block olarak kullanıldı
class ConvBlock(nn.Module):
def __init__(self, in_channels, out_channels):
super().__init__()
self.residual = nn.Sequential(
nn.Conv2d(in_channels, out_channels, kernel_size=3, stride=1, padding=1, bias=False),
nn.BatchNorm2d(out_channels),
nn.ReLU(inplace=True),
nn.Conv2d(out_channels, out_channels, kernel_size=3, stride=1, padding=1, bias=False),
nn.BatchNorm2d(out_channels),
nn.ReLU(inplace=True))
self.shortcut = nn.Sequential(
nn.Conv2d(in_channels, out_channels, kernel_size=1, stride=1),
nn.BatchNorm2d(out_channels),
nn.ReLU(inplace=True))
def forward(self, x):
return self.residual(x)+self.shortcut(x)
#U netin encoder mimarisi
class encoder_block(nn.Module):
#U-net encoder orjinal paper'den.
def __init__(self, chs=(3,64,128,256,512,1024)):
super().__init__()
self.enc_blocks = nn.ModuleList([ConvBlock(chs[i], chs[i+1]) for i in range(len(chs)-1)])
self.pool = nn.MaxPool2d((2,2))
def forward(self, x):
ftrs = []
for block in self.enc_blocks:
x = block(x)
ftrs.append(x)
x = self.pool(x)
return ftrs
#U netin decoder mimarisi
class decoder_block(nn.Module):
def __init__(self, chs=(1024,512, 256, 128, 64)):
super().__init__()
self.chs = chs
self.upconvs = nn.ModuleList([nn.ConvTranspose2d(chs[i], chs[i+1], 2, 2) for i in range(len(chs)-1)])
self.dec_blocks = nn.ModuleList([ConvBlock(chs[i], chs[i+1]) for i in range(len(chs)-1)])
def forward(self, x, encoder_features):
for i in range(len(self.chs)-1):
x = self.upconvs[i](x)
enc_ftrs = self.crop(encoder_features[i], x)
x = torch.cat([x, enc_ftrs], dim=1)
x = self.dec_blocks[i](x)
return x
def crop(self, enc_ftrs, x):
_, _, H, W = x.shape
enc_ftrs = torchvision.transforms.CenterCrop([H, W])(enc_ftrs)
return enc_ftrs
##Bbox tahmını ıcın encoderin cıktısına regressıon model baglanıldı.
class RegressionModel(nn.Module):
def __init__(self, input_size, hidden_size,hidden2,hidden3, output_size):
super(RegressionModel, self).__init__()
self.fc1 = nn.Linear(input_size, hidden_size)
self.relu = nn.ReLU()
self.fc2 = nn.Linear(hidden_size, hidden2)
self.fc3 = nn.Linear(hidden2, hidden3)
self.fc4 = nn.Linear(hidden3, output_size)
self.sigmoid = nn.Sigmoid()
def forward(self, x):
x = self.fc1(x)
x = self.relu(x)
x = self.fc2(x)
x = self.relu(x)
x = self.fc3(x)
x = self.relu(x)
x = self.fc4(x)
x= self.sigmoid(x)
return x
#modelin tamamlanması
class build_unet(nn.Module):
def __init__(self, enc_chs=(3,64,128,256,512,1024), dec_chs=(1024, 512, 256, 128, 64), num_class=1, retain_dim=False):
super().__init__()
self.encoder = encoder_block(enc_chs)
self.decoder = decoder_block(dec_chs)
self.regression = RegressionModel(1024,512,256,64,4)
self.head = nn.Conv2d(dec_chs[-1], num_class, 1)
#self.sigmoid = nn.Sigmoid()# Kullanılan loss'dan dolayı kullanılmasına gerek yok.
self.retain_dim = retain_dim
def forward(self, x):
enc_ftrs = self.encoder(x)
_x=enc_ftrs[::-1][0]
enc_features_flat = torch.mean(_x.view(_x.size(0), _x.size(1), -1), dim=2)
out_reg = self.regression(enc_features_flat)
out = self.decoder(enc_ftrs[::-1][0], enc_ftrs[::-1][1:])
out = self.head(out)
return out,out_reg
#4 VE 5.BÖLÜM
def train(BATCH_SIZE,NUM_EPOCHS,TEST_SPLIT,INIT_LR,INPUT_IMAGE_WIDTH,INPUT_IMAGE_HEIGHT,TEST_PATHS,MODEL_PATH_BEST,MODEL_PATH_LAST,PLOT_PATH,DEVICE,df1):
temp=0
global flag
global transforms
imagePaths =images_dataset
maskPaths = mask_dataset
#verileri %85 train %15 test olarak ayır.
X_train, X_test,y_train, y_test, bbox_train, bbox_test = train_test_split(os.listdir(imagePaths),os.listdir(maskPaths),df1.to_numpy(),
random_state=34,
test_size=TEST_SPLIT)
(trainImages, testImages)=X_train,X_test
(trainMasks, testMasks) =y_train,y_test
#verilerin full pathini veriyoruz
trainImages = list(map(lambda orig_string: imagePaths+"/"+orig_string , trainImages))
testImages = list(map(lambda orig_string: imagePaths+"/"+orig_string , testImages))
trainMasks = list(map(lambda orig_string: maskPaths+"/"+orig_string , trainMasks))
testMasks = list(map(lambda orig_string: maskPaths+"/"+orig_string , testMasks))
bbox_train=torch.tensor(bbox_train) # bboxları tensora çeviriyoruz
bbox_test=torch.tensor(bbox_test)
#
print("[INFO] saving testing image paths...")
f = open(TEST_PATHS, "w")
f.write("\n".join(testImages))
f.close()
#
# pytorch ıcın gereklı olan transformlar belırlenmıstır.
transforms = transforms.Compose([transforms.ToPILImage(),
transforms.Resize((INPUT_IMAGE_HEIGHT,
INPUT_IMAGE_WIDTH)),
transforms.ToTensor()])
# train ve test datasetlerın ayarlanması
trainDS = CustomDataset(imagePaths=trainImages, maskPaths=trainMasks,
transforms=transforms,bbox=bbox_train)
testDS = CustomDataset(imagePaths=testImages, maskPaths=testMasks,
transforms=transforms,bbox=bbox_test)
print(f"[INFO] found {len(trainDS)} examples in the training set...")
print(f"[INFO] found {len(testDS)} examples in the test set...")
# Training ve test veri setlerinin düzenlenmesi
trainLoader = DataLoader(trainDS, shuffle=False,
batch_size=BATCH_SIZE,
num_workers=os.cpu_count())
testLoader = DataLoader(testDS, shuffle=False,
batch_size=BATCH_SIZE,
num_workers=os.cpu_count())
unet = build_unet().to(DEVICE)
lossFunc = nn.BCEWithLogitsLoss() #for mask
opt = Adam(unet.parameters(), lr=INIT_LR)
lossFunc2= nn.L1Loss() #for bbox
trainSteps = len(trainDS) // BATCH_SIZE
testSteps = len(testDS) // BATCH_SIZE
H = {"train_loss": [], "test_loss": []}
print("[INFO] training the network...")
startTime = time.time()
#pytoch train
for e in tqdm(range(NUM_EPOCHS)):
#Eğitim için modeli train modunda başlatalım
unet.train()
totalTrainLoss = 0
totalTestLoss = 0
for (i, (x, y, labels)) in enumerate(trainLoader):
(x, y, labels) = (x.to(DEVICE), y.to(DEVICE),labels.to(DEVICE))
#images, mask, bounding boxes
pred,pred_bbox= unet(x)
loss = lossFunc(pred, y) #mask loss BCEWithLogitsLoss
loss2= lossFunc2(pred_bbox,labels)#bbox loss L1Loss
totalTrainLoss = (loss2 * 10) + loss #loss optimizasyonu için testler yapılmıştır. Bbox losu 10 ile çarpılmıştır.
opt.zero_grad()
totalTrainLoss.backward()
opt.step()
with torch.no_grad():
unet.eval() #test modunda modeli çağırıyoruz
for (i, (x, y, labels)) in enumerate(testLoader):
(x, y,labels) = (x.to(DEVICE), y.to(DEVICE),labels.to(DEVICE))
pred,pred_bbox= unet(x)
loss = lossFunc(pred, y) # mask loss
loss2= lossFunc2(pred_bbox,labels) #bbox loss
totalTestLoss = loss + (loss2*10)
avgTrainLoss = totalTrainLoss / trainSteps
avgTestLoss = totalTestLoss / testSteps
H["train_loss"].append(avgTrainLoss.cpu().detach().numpy())
H["test_loss"].append(avgTestLoss.cpu().detach().numpy())
# print the model training and validation information
print("[INFO] EPOCH: {}/{}".format(e + 1, NUM_EPOCHS))
print("Train loss: {:.6f}, Test loss: {:.4f}".format(
avgTrainLoss, avgTestLoss))
#En iyi modelin kaydedilmesi
if(temp>avgTestLoss.cpu().detach().numpy() or flag==1):
torch.save(unet,MODEL_PATH_BEST)
print("EN İYİ MODEL KAYDEDİLDİ")
flag=0
temp=avgTestLoss.cpu().detach().numpy()
###
endTime = time.time()
print("[INFO] total time taken to train the model: {:.2f}s".format(
endTime - startTime))
print("ALL LOSS",H)
### Eğitim loss değerlerini görselleştirme
plt.style.use("ggplot")
plt.figure()
plt.plot(H["train_loss"], label="train_loss")
plt.plot(H["test_loss"], label="test_loss")
plt.title("Training Loss on Dataset")
plt.xlabel("Epoch #")
plt.ylabel("Loss")
plt.legend(loc="lower left")
plt.savefig(PLOT_PATH)
###
#Modelin son halini kaydeder
torch.save(unet,MODEL_PATH_LAST)
print("SON MODEL KAYDEDİLDİ")
#### 1.B, 1.C
def create_dataframe(file,x,y,w,h):
global count
global df1
global r
path_list.append(mask_dataset+"/"+file)
arr[count][0]=float(x) # x_min
arr[count][1]=float(y) # y_min
arr[count][2]=float(x+w) # x_max
arr[count][3]=float(y+h) # y_max
count+=1
#pathlerden ve box lardan dataframe olusturur. df1'i kullanarak işlemlerimize devam edicez.
if(count==r):
df1=pd.DataFrame(arr, columns = ['x_min','y_min','x_max','y_max']) # bbox etiketi için kullanıcağız
df2=pd.DataFrame(path_list, columns = ['path'])
df=pd.concat([df1,df2],axis=1)
#print(df1.head())
#### 1.A
def mask_to_bbox(mask_dataset):
#Max Contour yöntemini kullanarak verilen maskelerin bounding boxeslarını çıkarır ve x_min, x_max, y_min, y_max olarak depolar.
for file in os.listdir(mask_dataset):
file=mask_dataset+"/"+file
img = cv2.imread(file)
img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
contours, _ = cv2.findContours(img, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)
if len(contours) != 0:
for c in contours:
c = max(contours, key = cv2.contourArea)
x,y,w,h = cv2.boundingRect(c)
create_dataframe(file,x,y,w,h)
if __name__ == "__main__":
#args
### 1.D
parser = argparse.ArgumentParser(description='-')
parser.add_argument('--batch_size', type=int, default=32,
help='Batch size değeri (varsayılan: 8)')
parser.add_argument('--num_epochs', type=int, default=200,
help='num_epochs size değeri (varsayılan: 200)')
parser.add_argument('--lr', type=float, default=0.001,
help='learnig_rate değeri (varsayılan: 0.001)')
parser.add_argument('--test_split', type=float, default=0.15,
help='Test split değeri değeri (varsayılan: 0.15)')
parser.add_argument('--BASE_OUTPUT', type=str, default="output",
help='Outputların kaydedileceği konum(varsayılan: ./output)')
parser.add_argument('--INPUT_IMAGE_WIDTH', type=int, default=128,
help='INPUT_IMAGE_WIDTH(varsayılan: 128)')
parser.add_argument('--INPUT_IMAGE_HEIGHT', type=int, default=128,
help='INPUT_IMAGE_HEIGHT(varsayılan: 128)')
args = parser.parse_args()
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
print("Device",DEVICE)
BATCH_SIZE,NUM_EPOCHS,TEST_SPLIT,INIT_LR,INPUT_IMAGE_WIDTH,INPUT_IMAGE_HEIGHT,TEST_PATHS,MODEL_PATH_BEST,MODEL_PATH_LAST,PLOT_PATH=args_init(args)
mask_to_bbox(mask_dataset)
train(BATCH_SIZE,NUM_EPOCHS,TEST_SPLIT,INIT_LR,INPUT_IMAGE_WIDTH,INPUT_IMAGE_HEIGHT,TEST_PATHS,MODEL_PATH_BEST,MODEL_PATH_LAST,PLOT_PATH,DEVICE,df1)