-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmain.py
242 lines (206 loc) · 9.74 KB
/
main.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
import torch
import torch.utils.data as data
from utils import *
import time
import numpy as np
torch.set_printoptions(precision=3, edgeitems=14, linewidth=350)
os.environ["CUDA_VISIBLE_DEVICES"]="2"
from timm.loss import *
from model import *
from casmeii import *
from sklearn.metrics.pairwise import cosine_similarity
sys.stdout = Logger("logs/your_log.log")
def run_training(cut_alpha=1.0,mix_alpha=1.0,num_classes=5):
#writer = SummaryWriter('./tensor_log')
args = parse_args()
##data normalization for both training set
criterion = SoftTargetCrossEntropy()#torch.nn.CrossEntropyLoss()
criterion_val=torch.nn.CrossEntropyLoss()
data_aug = transforms.Compose([
transforms.RandomHorizontalFlip(p=0.5),
transforms.RandomRotation(4),
transforms.RandomCrop(224, padding=4),
])
# three classes
# mean：[0.303,0.366,0.509]
# std:[0.119,0.135,0.187]
# mean;[0.016,0.003]
# std:[0.994,0.919]
# five classes
# mean:[0.296,0.358,0.500]
# std:[0.123,0.136,0.185]
# mean:[0.006,0.066]
# std:[0.956,0.835]
flow_normal = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize(mean=[0.006, 0.066],
std=[0.956, 0.833]),
])
onset_normal = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize(mean=[0.296, 0.358, 0.500],
std=[0.123, 0.136, 0.185]),
])
# leave one subject out protocal
LOSO = ['17', '26', '16', '9', '5', '24', '2', '13', '4', '23', '11', '12', '8', '14', '3',
'19', '1','18',
'10','20', '21', '22', '15', '6', '25', '7']
val_now = 0
num_sum = 0
pos_pred_ALL = torch.zeros(5)
pos_label_ALL = torch.zeros(5)
TP_ALL = torch.zeros(5)
acc_list = []
epoch_list = []
norm_transforms = transforms.Compose([
transforms.ToTensor(),
])
for subj in LOSO:
train_dataset = RafDataSet(args.raf_path, phase='train', num_loso=subj, transform_flow=flow_normal,transform_onset=onset_normal,transform_aug=data_aug,num_classes=num_classes)
val_dataset = RafDataSet(args.raf_path, phase='test', num_loso=subj, transform_flow=flow_normal,transform_onset=onset_normal,num_classes=num_classes)
train_loader = torch.utils.data.DataLoader(train_dataset,
batch_size=34,
num_workers=args.workers,
shuffle=True,
pin_memory=True)
val_loader = torch.utils.data.DataLoader(val_dataset,
batch_size=34,
num_workers=args.workers,
shuffle=False,
pin_memory=True)
print('num_sub', subj)
print('Train set size:', train_dataset.__len__())
print('Validation set size:', val_dataset.__len__())
max_epoch = 0
max_corr = 0
max_f1 = 0
max_pos_pred = torch.zeros(5)
max_pos_label = torch.zeros(5)
max_TP = torch.zeros(5)
net_all = MixMe_Net(pretrained=True, num_classes=num_classes, return_attn=True,merge=True)
optimizer_all = torch.optim.Adam(net_all.parameters(),lr=3e-5)
net_all = net_all.cuda()
for i in range(1, 100):
running_loss = 0.0
correct_sum = 0
iter_cnt = 0
net_all.train()
for batch_i, (
flow,onset, label_all) in enumerate(train_loader):
iter_cnt += 1
label_all=label_all.cuda()
onset=onset.cuda()
flow = flow.cuda()
mixup_lam = np.random.beta(mix_alpha, mix_alpha)
mask,lam=generate_flow_mask(flow,lam_alpha=cut_alpha,type='avg')
mask_224=torch.nn.functional.interpolate(mask,size=(224,224),mode='nearest')
index = torch.randperm(flow.size()[0]).cuda()
label1=label_all
label2=label_all[index]
mix_flow = flow*(mask_224) + mixup_lam * flow * (1 - mask_224) + (1 - mixup_lam) * flow[index] * (
1 - mask_224)
ALL, attn, final = net_all(mix_flow, onset)
label_mix=label_mix = transmix_label(label1, label2, mask, attn, mixup_lam, num_classes=num_classes,
smoothing=0.0)
loss_all = criterion(ALL,label_mix)#loss_func(criterion,ALL_mix)
#writer.add_scalar('trans_loss_mixup/'+subj, loss_all, i)
optimizer_all.zero_grad()
loss_all.backward()
optimizer_all.step()
running_loss += loss_all
_, predicts = torch.max(ALL, 1)
correct_num = torch.eq(predicts, label_all).sum()
correct_sum += correct_num
if i >= 0:
acc = correct_sum.float() / float(train_dataset.__len__())
running_loss = running_loss / iter_cnt
print('[Epoch %d] Training Loss: %.3f' % (i, running_loss))
pos_label = torch.zeros(5)
pos_pred = torch.zeros(5)
TP = torch.zeros(5)
with torch.no_grad():
running_loss = 0.0
iter_cnt = 0
bingo_cnt = 0
sample_cnt = 0
net_all.eval()
for batch_i, ( flow,onset,label_all) in enumerate(val_loader):
label_all = label_all.cuda()
flow = flow.cuda()
onset=onset.cuda()
##test
(ALL,attn,_) = net_all(flow,onset)
loss = criterion_val(ALL, label_all)
running_loss += loss
iter_cnt += 1
_, predicts = torch.max(ALL, 1)
correct_num = torch.eq(predicts, label_all)
bingo_cnt += correct_num.sum().cpu()
sample_cnt += ALL.size(0)
for cls in range(5):
for element in predicts:
if element == cls:
pos_label[cls] = pos_label[cls] + 1
for element in label_all:
if element == cls:
pos_pred[cls] = pos_pred[cls] + 1
for elementp, elementl in zip(predicts, label_all):
if elementp == elementl and elementp == cls:
TP[cls] = TP[cls] + 1
count = 0
SUM_F1 = 0
for index in range(5):
if pos_label[index] != 0 or pos_pred[index] != 0:
count = count + 1
SUM_F1 = SUM_F1 + 2 * TP[index] / (pos_pred[index] + pos_label[index])
AVG_F1 = SUM_F1 / count
running_loss = running_loss / iter_cnt
acc = bingo_cnt.float() / float(sample_cnt)
acc = np.around(acc.numpy(), 4)
#writer.add_scalar('trans_acc_mixup/' + subj, acc, i)
if bingo_cnt > max_corr:
max_corr = bingo_cnt
max_epoch = i
if AVG_F1 >= max_f1:
max_f1 = AVG_F1
max_pos_label = pos_label
max_pos_pred = pos_pred
max_TP = TP
print("[Epoch %d] Validation accuracy:%.4f. Loss:%.3f, F1-score:%.3f" % (i, acc, running_loss, AVG_F1))
if acc==1.:
print('achieve 100%acc, break')
break
num_sum = num_sum + max_corr
pos_label_ALL = pos_label_ALL + max_pos_label
pos_pred_ALL = pos_pred_ALL + max_pos_pred
TP_ALL = TP_ALL + max_TP
count = 0
SUM_F1 = 0
for index in range(5):
if pos_label_ALL[index] != 0 or pos_pred_ALL[index] != 0:
count = count + 1
SUM_F1 = SUM_F1 + 2 * TP_ALL[index] / (pos_pred_ALL[index] + pos_label_ALL[index])
F1_ALL = SUM_F1 / count
val_now = val_now + val_dataset.__len__()
acc_now=(max_corr/val_dataset.__len__()).item()
#writer.add_scalar('trans_acc', acc_now, subj)
print("[subject %s] correct_num:%d sum:%d ACC: %.4f " % (subj, max_corr, val_dataset.__len__(),acc_now))
print("[ALL_corr]: %d [ALL_val]: %d [ALL_ACC]:%.4f" % (int(num_sum), int(val_now), num_sum / val_now))
print("[F1_now]: %.4f [F1_ALL]: %.4f" % (max_f1, F1_ALL))
print('max_epoch:', str(max_epoch))
acc_list.append(round(acc_now,4))
epoch_list.append(str(max_epoch))
print("--------------------------------------------------------------------------------------------------")
print('cut alpha,', cut_alpha)
print('mix alpha', mix_alpha)
print("Acc_list:", acc_list)
print('max epoches:', epoch_list)
print("Total acc: %.4f"% (num_sum * 1.0 / val_now))
if __name__ == "__main__":
seed=2
start_time = time.time()
seed_torch(seed)
run_training(cut_alpha=2.0, mix_alpha=1.0,num_classes=5)
end_time = time.time()
print("start time:", time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(start_time)))
print("end_time:", time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(end_time)))