/
FaceClassifier.py
137 lines (119 loc) · 4.63 KB
/
FaceClassifier.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
import torch
import torch.nn as nn
import numpy as np
import os
from torch.autograd import Variable
from MultiLabelDataLoaderAug import *
from torchvision import datasets, models, transforms
from HRModel import *
import torch.optim as optim
import torch.nn.functional as F
from os.path import exists, join, basename, dirname
from torch.optim import lr_scheduler
import math
# pytorch0.4, pytorch1.0
My_CLASSES = ( # always index 0
'null',
'face'
)
codeDict={
'del':'null',
'FaceCrop':'face'
}
FilePre='test_HRNet_Face'
torch.cuda.set_device(0)
# Model
print('==> Building model..')
MyClassNum = len(My_CLASSES)
model = MultiLabelClassifierHRNetW30(MyClassNum)
#model = torch.nn.DataParallel(model,device_ids=[0,2,4,6]).cuda()
model.cuda()
m=nn.Sigmoid()
init_lr = 1e-2
img_size=256
optimizer = optim.SGD(model.parameters(), lr= init_lr, momentum=0.5)
def transform_train(img, labels):
img = random_noise(img)
img,_ = resize(img, boxes=None, size=(img_size,img_size), random_interpolation=True)
img = random_flip(img)
img = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.485,0.456,0.406),(0.229,0.224,0.225))
])(img)
return img, labels
trainset = ListDatasetLite(root='/home/wynmew/data/tachie',
list_file='/home/wynmew/data/tachie/faceCropList', # dataset with neg
classes = My_CLASSES,
barcode_Dict = codeDict,
transform=transform_train)
trainBatchSize = 12
trainloader = torch.utils.data.DataLoader(trainset, batch_size=trainBatchSize, shuffle=True, drop_last=True, num_workers=0)
def addlog(x):
with open(FilePre+"_Log","a+") as outfile:
outfile.write(x + "\n")
k=1 #OHEM
def train(epoch, model, loss_fn, optimizer, dataloader,log_interval=50):
model.train()
train_loss = 0
for batch_idx, (inputs, cls_targets) in enumerate(dataloader):
inputs = Variable(inputs.cuda())
optimizer.zero_grad()
labelsPre = model(inputs)
lossCLS = Variable(torch.FloatTensor(1)).zero_()
lossCLS = lossCLS.cuda()
# calculate loss, OHEM
for bID in range(trainBatchSize): # batch itr
tmpBatchLoss = Variable(torch.FloatTensor(1)).zero_()
tmpBatchLoss = tmpBatchLoss.cuda()
for cID in range(MyClassNum): # class itr
sub_loss = loss_fn(m(labelsPre[cID][bID]), Variable(cls_targets[bID][cID].cuda()))
tmpBatchLoss += sub_loss
lossCLS = torch.cat((lossCLS, tmpBatchLoss), 0)
index = torch.topk(lossCLS, int(k * lossCLS.size()[0]))
valid_lossCLS = lossCLS[index[1]]
lossCLS=torch.mean(valid_lossCLS)
loss = lossCLS
loss.backward()
optimizer.step()
#train_loss += loss.data.cpu().numpy()[0] #pytorch0.3
train_loss += loss.data.cpu().numpy() #pytorch1.0
if batch_idx % log_interval == 0:
print('Train Epoch: {} [{}/{} ({:.0f}%)] Loss: {:.6f}'.format(
epoch, batch_idx , len(dataloader), 100. * batch_idx / len(dataloader), loss.data)) #pytorch1.0
line = "Train Epoch: " + str(epoch) + " " + str(100. * batch_idx / len(dataloader)) + " " + str(loss.data) #pytorch1.0
#print('Train Epoch: {} [{}/{} ({:.0f}%)] Loss: {:.6f}'.format(
# epoch, batch_idx , len(dataloader), 100. * batch_idx / len(dataloader), loss.data[0])) # pytorch0.3
#line = "Train Epoch: " + str(epoch) + " " + str(100. * batch_idx / len(dataloader)) + " " + str(loss.data[0])# pytorch0.3
addlog(line)
train_loss /= len(dataloader)
print('Train set: Average loss: {:.4f}'.format(train_loss))
line = "Train set: Average loss: " + str(train_loss)
#add(line)
return train_loss
def adjust_lr(optimizer, epoch, maxepoch, init_lr, power = 0.9):
lr = init_lr * (1-epoch/maxepoch)**power
for param_group in optimizer.param_groups:
param_group['lr'] = lr
return lr
cwd = os.getcwd()
print(cwd)
lossFun = nn.MSELoss()
best_test_loss = float("inf")
print('Starting training...')
start_epoch = 1
end_epoch = 2
for epoch in range(start_epoch, end_epoch + 1):
train_loss = train(epoch, model, lossFun, optimizer, trainloader, log_interval=1)
for param_group in optimizer.param_groups:
print(param_group['lr'])
lr_now = adjust_lr(optimizer, epoch, end_epoch + 1, init_lr, power=10)
print(lr_now)
line = "lr_Now: " + str(lr_now)
print('Saving..')
state = {
'net': model.state_dict(),
#'net': model.module.state_dict(), #DataParallel
'epoch': epoch,
}
torch.save(state, FilePre+str(epoch)+'.pth.tar')
print('Done!')