-
Notifications
You must be signed in to change notification settings - Fork 0
/
cl.py
157 lines (140 loc) · 5.93 KB
/
cl.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
import chainer
from chainer import links as L
from chainer import functions as F
import numpy as np
import argparse
from chainer import serializers
import network
from chainer import training
from chainer.training import extension
from chainer.training import extensions
from updater_cl import Updater
from iterator import MyIterator
from encoder import InceptionResNetV2
from util import MultistepShift
#from config import Config
class Config(object):
celebA_dir = '/root/share/datasets/CelebA/'
ms_celeb_dir='/root/share/datasets/MSCeleb1M/raw'
#pretrainmodel_path='inception_resnet.npz'
#pretrainmodel_path='result/Classification_40000.npz'
pretrainmodel_path='/root/inception_resnet.npz'
display_interval=10
lr=0.01
gpu=6
batchsize=32*3
max_iter= 1184160/batchsize*30
epoch_iter= 1184160/batchsize
#max_iter= 162770/batchsize*30
evaluation_interval=100
out='msceleb_pretrain3'
display_interval=10
snapshot_interval=50000
def eval_classification(dataset, cl, evaln, batchn, gpu,msg='tmp'):
def load_image(dataset, i,i_end, random_crop=False):
imgs=[]; imgs_out=[]; labels=[]
crop_size=235
for i in range(i,i_end):
imn, label=dataset[i]
img=cv2.imread(imn)
img=img[:,:,::-1]
img_in=cv2.resize(img,(crop_size,crop_size)).astype(np.float32)
img_in /= 255.
img_in -= 0.5 # mean 0.5
img_in *= 2.0 # std 0.5
img_in=img_in.transpose(2,0,1).astype(np.float32)
imgs.append(img_in)
labels.append(label)
return (np.asarray(imgs), labels)
def get_acc(out,label,topk=1):
sortid=np.argsort(-out)
correctn=np.sum(sortid[:,0:topk]==label)
topk_acc=float(correctn)/(len(out))
return topk_acc
@chainer.training.make_extension()
def evaluation(trainer):
xp = cl.xp
mean_top10_acc1=0
mean_top100_acc1=0
mean_top1000_acc1=0
for i in range(evaln):
start=random.randint(0,100)
start=min(len(dataset)-batchn,start)
end=start+batchn
batch, label = load_image(dataset,start,end)
batchsize = len(batch)
x = [];
for i in range(batchsize):
x.append(np.asarray(batch[i]).astype("f"))
with cupy.cuda.Device(gpu):
x_in = Variable(xp.asarray(x))
with chainer.using_config('train', False), chainer.using_config('enable_backprop', False):
y = cl(x_in)
y=F.softmax(y)
label = np.asarray(label, dtype=np.int32)
label = np.reshape(label, (label.shape[0],1))
#top1_acc = get_acc(chainer.cuda.to_cpu(out.data),label,topk=1)
top10_acc1 = get_acc(chainer.cuda.to_cpu(y.data),label,topk=10)
top100_acc1 = get_acc(chainer.cuda.to_cpu(y.data),label,topk=100)
top1000_acc1 = get_acc(chainer.cuda.to_cpu(y.data),label,topk=1000)
#mean_top1_acc += top1_acc/batchn
mean_top10_acc1 += top10_acc1
mean_top100_acc1 += top100_acc1
mean_top1000_acc1 += top1000_acc1
mean_top10_acc1 /= evaln
mean_top100_acc1 /= evaln
mean_top1000_acc1 /= evaln
print('y',msg)
print('top10score: ',mean_top10_acc1)
print('top100score: ',mean_top100_acc1)
print('top1000score: ',mean_top1000_acc1)
return evaluation
class Classification(chainer.Chain):
def __init__(self, encoder, identityn):
super(Classification, self).__init__()
with self.init_scope():
self.encoder = InceptionResNetV2()
self.last_linear = L.Linear(None, identityn)
def __call__(self, x):
x=self.encoder(x)
x = F.average_pooling_2d(x, (x.shape[2],x.shape[3]))
x=self.last_linear(x)
return x
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument("--batch_size", default=16, type=int)
parser.add_argument("--wt_dec", default=5e-4, type=float)
parser.add_argument('--gpu', '-g', type=int, default=0, help='GPU ID (negative value indicates CPU)')
parser.add_argument('--lr', '-l', type=int, default=1e-3, help='learning rate')
parser.add_argument('--out', '-o', default='result', help='Directory to output the result')
args = parser.parse_args()
config=Config()
train_dataset, _, _, identityn = celebA_load(config)
train_iter = MyIterator(config, train_dataset, args.batch_size)
opts = {}
encoder=InceptionResNetV2()
serializers.load_npz(config.pretrainmodel_path, encoder)
clmodel=Classification(encoder, identityn).to_gpu(config.gpu)
models = clmodel
updater_args = {
"iterator": {'main': train_iter},
"device": args.gpu
}
opts["opt"] = make_sgd_optimizer(clmodel, config.lr)
updater_args["optimizer"] = opts
updater_args["models"] = models
updater_args["config"] = config
updater = Updater(**updater_args)
report_keys = ["loss"]
trainer = training.Trainer(updater, (config.max_iter, 'iteration'), out=config.out)
trainer.extend(MultistepShift('lr', 0.1, [config.epoch_iter*10,config.epoch_iter*20], 1e-2, optimizer=opts["opt"]))
trainer.extend(extensions.snapshot_object(
clmodel, clmodel.__class__.__name__ + '_{.updater.iteration}.npz'), trigger=(config.snapshot_interval, 'iteration'))
trainer.extend(extensions.LogReport(keys=report_keys,
trigger=(config.display_interval, 'iteration')))
trainer.extend(extensions.PrintReport(report_keys), trigger=(config.display_interval, 'iteration'))
trainer.extend(extensions.ProgressBar(update_interval=10))
trainer.extend(eval_classification(train_dataset, clmodel, 1, 100, config.gpu,msg='train'), trigger=(config.evaluation_interval, 'iteration'),
priority=extension.PRIORITY_WRITER)
# Run the training
trainer.run()