-
Notifications
You must be signed in to change notification settings - Fork 0
/
main_contrast.py
314 lines (281 loc) · 14.1 KB
/
main_contrast.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
import os, sys, datetime, time, random, argparse, copy
import numpy as np
import scipy.io as scio
# os.environ["CUDA_VISIBLE_DEVICES"] = "3"
import torch
import torch.optim as optim
import torch.nn as nn
import torchvision
from torchvision import transforms
from Datasets import MEGC2019_SI_useNP as MEGC2019
import LossFunctions
import Metrics as metrics
# from model.DualATTNet_womad import DualATTNet
from model.DualATTNet import DualATTNet
def arg_process():
parser = argparse.ArgumentParser()
parser.add_argument('--dataversion', type=str, default='cde_flow_np', help='the version of input data') #cde_flow和cde_flow_puzzle
parser.add_argument('--epochs', type=int, default=30, help='the number of training epochs')
parser.add_argument('--gpuid', default='cuda:0', help='the gpu index for training')
parser.add_argument('--learningrate', type=float, default=0.001, help='the learning rate for training')
parser.add_argument('--modelname', default='DualATTNet', help='the model architecture') # res18 rcn_a meroi
parser.add_argument('--batchsize', type=int, default=128, help='the batch size for training')
parser.add_argument('--lossfunction', default='crossentropy', help='the loss functions') # crossentropy, focal, balanced
parser.add_argument('--optimizer', default='adamW', help='the optimizer') # adam,sgd
args = parser.parse_args()
return args
def train_model(model, dataloaders, cosine_sim, criterion, optimizer, device='cpu', num_epochs=25):
since = time.time()
# best_model_wts = copy.deepcopy(model.state_dict())
best_acc = 0.0
# scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=10, gamma=0.1)
for epoch in range(num_epochs):
print('\tEpoch {}/{}'.format(epoch, num_epochs - 1))
print('\t' + '-' * 10)
# Each epoch has a training
model.train() # Set model to training mode
running_loss = 0.0
running_corrects = 0
# Iterate over data
for j, samples in enumerate(dataloaders):
inputs1, class_labels = samples["data"], samples["class_label"]
inputs1 = torch.FloatTensor(inputs1).to(device)
class_labels = class_labels.to(device)
cosine_sim = cosine_sim.to(device)
# zero the parameter gradients
optimizer.zero_grad()
# forward to get model outputs and calculate loss
out_x1, out_x2, output_class = model(inputs1)
loss_contrast = 1 - cosine_sim(out_x1, out_x2)
loss_contrast = loss_contrast.mean()
loss_class = criterion(output_class, class_labels) # 计算output_class的loss
loss = 0.1 * loss_contrast + 1 * loss_class # 到时候con的系数可以调
# backward
loss.backward()
optimizer.step()
# scheduler.step()
# statistics
_, predicted = torch.max(output_class.data, 1)
running_loss += loss.item() * inputs1.size(0)
running_corrects += torch.sum(predicted == class_labels)
epoch_loss = running_loss / len(dataloaders.dataset)
epoch_acc = running_corrects.double() / len(dataloaders.dataset)
print('\t{} Loss: {:.4f} Acc: {:.4f}'.format('Train', epoch_loss, epoch_acc))
time_elapsed = time.time() - since
print('\tTraining complete in {:.0f}m {:.0f}s'.format(time_elapsed // 60, time_elapsed % 60))
return model
def test_model(model, dataloaders, device):
model.eval()
num_samples = len(dataloaders.dataset)
predVec = torch.zeros(num_samples)
labelVec = torch.zeros(num_samples)
start_idx = 0
end_idx = 0
for j, samples in enumerate(dataloaders):
inputs1, class_labels = samples["data"], samples["class_label"]
inputs1 = torch.FloatTensor(inputs1).to(device)
# update the index of ending point
end_idx = start_idx + inputs1.shape[0]
output_class = model(inputs1)
# _, predicted = torch.max(output_class.data, 1)
_, predicted = torch.max(output_class[2].data, 1)
predVec[start_idx:end_idx] = predicted
labelVec[start_idx:end_idx] = class_labels
# update the starting point
start_idx += inputs1.shape[0]
return predVec, labelVec
def main():
"""
Goal: process images by file lists, evaluating the datasize with different model size
Version: 5.0
"""
print('PyTorch Version: ', torch.__version__)
print('Torchvision Version: ', torchvision.__version__)
now = datetime.datetime.now()
random.seed(1)
torch.manual_seed(1)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
args = arg_process()
runFileName = sys.argv[0].split('.')[0]
# need to modify according to the enviroment
version = args.dataversion
gpuid = args.gpuid
model_name = args.modelname
num_epochs = args.epochs
lr = args.learningrate
batch_size = args.batchsize
loss_function = args.lossfunction
optimizer = args.optimizer
classes = 3
logPath = os.path.join('result', model_name + '_3ls_log_cde2019_head0.3.txt')
# logPath = os.path.join('result', runFileName + '_log_' + 'v{}'.format(args.dataversion) + '.txt')
# logPath = os.path.join('result', runFileName+'_log_'+'v{}'.format(version)+'.txt')
# resultPath = os.path.join('result', 'result_'+'v{}'.format(version)+'.pt')
data_transforms_train = transforms.Compose([
transforms.Resize(28),
# transforms.ToPILImage(),
transforms.RandomHorizontalFlip(0.5),
transforms.ToTensor(),
transforms.Normalize([0.5,0.5,0.5], [0.5,0.5,0.5]),
])
data_transforms_test = transforms.Compose([
transforms.Resize(28),
transforms.ToTensor(),
transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5]),
])
# move to GPU
print(torch.cuda.is_available())
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
# obtian the subject information in LOSO
verFolder = 'v_{}'.format(version)
filePath = os.path.join('data', 'MEGC2019', verFolder, 'subName.txt')
subjects = []
with open(filePath, 'r') as f:
for textline in f:
texts = textline.strip('\n')
subjects.append(texts)
# predicted and label vectors
preds_db = {}
preds_db['casme2'] = torch.tensor([])
preds_db['smic'] = torch.tensor([])
preds_db['samm'] = torch.tensor([])
preds_db['all'] = torch.tensor([])
labels_db = {}
labels_db['casme2'] = torch.tensor([])
labels_db['smic'] = torch.tensor([])
labels_db['samm'] = torch.tensor([])
labels_db['all'] = torch.tensor([])
# open the log file and begin to record
log_f = open(logPath, 'a')
log_f.write('{}\n'.format(now))
log_f.write('-' * 80 + '\n')
log_f.write('-' * 80 + '\n')
log_f.write('Results:\n')
allRST = []
time_s = time.time()
for subject in subjects:
print('Subject Name: {}'.format(subject))
print('---------------------------')
# random.seed(1)
# setup a dataloader for training
imgDir = os.path.join('data', 'MEGC2019', verFolder, '{}_train.txt'.format(subject))
print(imgDir)
image_db_train = MEGC2019(imgList=imgDir, transform=data_transforms_train)
dataloader_train = torch.utils.data.DataLoader(image_db_train, batch_size=batch_size, shuffle=True,
num_workers=1)
# Initialize the model
print('\tCreating deep model....')
if model_name == 'DualATTNet':
model_ft = DualATTNet()
params_to_update = model_ft.parameters()
if optimizer == 'sgd':
optimizer_ft = optim.SGD(params_to_update, lr=lr, momentum=0.9)
elif optimizer == 'adam':
optimizer_ft = optim.Adam(params_to_update, lr=lr)
elif optimizer == 'adamW':
optimizer_ft = optim.AdamW(params_to_update, lr=lr)
if loss_function == 'crossentropy':
criterion = nn.CrossEntropyLoss()
# 若wetghtedCE loss
# class_weight = torch.FloatTensor([0.2857, 0.1329, 0.1695]).cuda(args.gpuid) # half-oversampling [1.0, 2.2936, 3.0120]和[0.2857, 0.1329,0.1695]
# criterion = nn.CrossEntropyLoss(weight=class_weight).cuda(args.gpuid)
elif loss_function == 'focal':
criterion = LossFunctions.FocalLoss(class_num=classes, device=device)
elif loss_function == 'balanced':
criterion = LossFunctions.BalancedLoss(class_num=classes, device=device)
elif loss_function == 'cosine':
criterion = LossFunctions.CosineLoss()
cosine_sim = torch.nn.CosineSimilarity(dim=1)
model_ft = model_ft.to(device) # from cpu to gpu
# Train and evaluate
model_ft = train_model(model_ft, dataloader_train, cosine_sim, criterion, optimizer_ft, device, num_epochs=num_epochs)
torch.save(model_ft, os.path.join('data', 'model_path', 'model_{}.pth').format(subject))
# Test model
imgDir = os.path.join('data', 'MEGC2019', verFolder, '{}_test.txt'.format(subject))
image_db_test = MEGC2019(imgList=imgDir, transform=data_transforms_test)
dataloaders_test = torch.utils.data.DataLoader(image_db_test, batch_size=batch_size, shuffle=False,
num_workers=1)
preds, labels = test_model(model_ft, dataloaders_test, device)
preds_np = np.array(preds)
labels_np = np.array(labels)
allRST.append([preds_np, labels_np])
acc = torch.sum(preds == labels).double() / len(preds)
print('\tSubject {} has the accuracy:{:.4f}\n'.format(subject, acc))
print('---------------------------\n')
log_f.write('\tSubject {} has the accuracy:{:.4f}\n'.format(subject, acc))
# saving the subject results
preds_db['all'] = torch.cat((preds_db['all'], preds), 0)
labels_db['all'] = torch.cat((labels_db['all'], labels), 0)
if subject.find('sub') != -1:
preds_db['casme2'] = torch.cat((preds_db['casme2'], preds), 0)
labels_db['casme2'] = torch.cat((labels_db['casme2'], labels), 0)
else:
if subject.find('s') != -1:
preds_db['smic'] = torch.cat((preds_db['smic'], preds), 0)
labels_db['smic'] = torch.cat((labels_db['smic'], labels), 0)
else:
preds_db['samm'] = torch.cat((preds_db['samm'], preds), 0)
labels_db['samm'] = torch.cat((labels_db['samm'], labels), 0)
time_e = time.time()
hours, rem = divmod(time_e - time_s, 3600)
miniutes, seconds = divmod(rem, 60)
# evaluate all data
allRST = np.array(allRST)
rstPath = os.path.join('result', model_name + '_lr' + str(lr) + '_rst_head0.3.mat')
scio.savemat(rstPath, {'Xia': allRST})
eval_acc = metrics.accuracy()
eval_f1 = metrics.f1score()
acc_w, acc_uw = eval_acc.eval(preds_db['all'], labels_db['all'])
f1_w, f1_uw = eval_f1.eval(preds_db['all'], labels_db['all'])
print('\nThe dataset has the ACC:{:.4f} and UAR and UF1:{:.4f} and {:.4f}'.format(acc_w, acc_uw, f1_uw))
log_f.write(
'\nOverall:\n\tthe ACC:{:.4f} and the UAR and UF1 of all data are {:.4f} and {:.4f}\n'.format(acc_w, acc_uw,
f1_uw))
# casme2
if preds_db['casme2'].nelement() != 0:
acc_w, acc_uw = eval_acc.eval(preds_db['casme2'], labels_db['casme2'])
f1_w, f1_uw = eval_f1.eval(preds_db['casme2'], labels_db['casme2'])
print('\nThe casme2 dataset has the ACC:{:.4f} and UAR and UF1:{:.4f} and {:.4f}'.format(acc_w, acc_uw, f1_uw))
log_f.write('\tthe ACC:{:.4f} and UAR and UF1 of casme2 are {:.4f} and {:.4f}\n'.format(acc_w, acc_uw, f1_uw))
# smic
if preds_db['smic'].nelement() != 0:
acc_w, acc_uw = eval_acc.eval(preds_db['smic'], labels_db['smic'])
f1_w, f1_uw = eval_f1.eval(preds_db['smic'], labels_db['smic'])
print('\nThe smic dataset has the ACC:{:.4f} and UAR and UF1:{:.4f} and {:.4f}'.format(acc_w, acc_uw, f1_uw))
log_f.write('\tthe ACC:{:.4f} and UAR and UF1 of smic are {:.4f} and {:.4f}\n'.format(acc_w, acc_uw, f1_uw))
# samm
if preds_db['samm'].nelement() != 0:
acc_w, acc_uw = eval_acc.eval(preds_db['samm'], labels_db['samm'])
f1_w, f1_uw = eval_f1.eval(preds_db['samm'], labels_db['samm'])
print('\nThe samm dataset has the ACC:{:.4f} and UAR and UF1:{:.4f} and {:.4f}'.format(acc_w, acc_uw, f1_uw))
log_f.write('\tthe ACC:{:.4f} and UAR and UF1 of samm are {:.4f} and {:.4f}\n'.format(acc_w, acc_uw, f1_uw))
# writing parameters into log file
print('\tNetname:{}, Dataversion:{}\n\tLearning rate:{}, Epochs:{}, Batchsize:{}.'.format(model_name, version, lr,
num_epochs, batch_size))
print('\tElapsed time: {:0>2}:{:0>2}:{:05.2f}'.format(int(hours), int(miniutes), seconds))
# log_f.write(
# '\nOverall:\n\tthe weighted and unweighted accuracy of all data are {:.4f} and {:.4f}\n'.format(acc_w, acc_uw))
log_f.write(
'\nSetting:\tNetname:{}, Dataversion:{}\n\tLearning rate:{}, Epochs:{}, Batchsize:{}.\n'.format(model_name,
version,
lr,
num_epochs,
batch_size))
# # save subject's results
# torch.save({
# 'predicts':preds_db,
# 'labels':labels_db,
# 'weight_acc':acc_w,
# 'unweight_acc':acc_uw
# },resultPath)
log_f.write('-' * 80 + '\n')
log_f.write('-' * 80 + '\n')
log_f.write('\n')
log_f.close()
if __name__ == '__main__':
# os.environ["CUDA_VISIBLE_DEVICES"] = "3"
# main()
for i in range(10): # 执行main函数10次
print('第{}次'.format(i+1))
main()