-
Notifications
You must be signed in to change notification settings - Fork 6
/
train_resnext_ibn_a_w_d_v2.py
231 lines (224 loc) · 13.4 KB
/
train_resnext_ibn_a_w_d_v2.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
"""
We modified resnext architecture as illustrated in following paper
Jie Hu, Li Shen, Gang Sun. "Squeeze-and-Excitation Networks" https://arxiv.org/pdf/1709.01507v1.pdf
(a) The first 7x7 convoluational layer was replaced with three consecutive 3x3 convolutional layers.
(b) The down-sampling projection 1x1 with stride-2 convolution was replaced with a 3x3 stride-2 convolution to preserve information.
(c) A dropout layer (with a drop ratio of 0.2) was inserted before the classifier layer to prevent overfitting.
(d) Label-smoothing regularization (as introduced in Christian Szegedy et. al. "Rethinking the Inception Architecture for Computer Vision") was used during training.
So we called it as resnext_w_d_v2
Modified by Lin Xiong Sep-13, 2017
"""
import argparse,logging,os
import mxnet as mx
from symbol_resnext_ibn_a_w_d_v2 import resnext
logger = logging.getLogger()
logger.setLevel(logging.INFO)
formatter = logging.Formatter('%(asctime)s - %(message)s')
console = logging.StreamHandler()
console.setFormatter(formatter)
logger.addHandler(console)
def multi_factor_scheduler(begin_epoch, epoch_size, step=[30, 60, 90, 95, 115, 120], factor=0.1):
step_ = [epoch_size * (x-begin_epoch) for x in step if x-begin_epoch > 0]
return mx.lr_scheduler.MultiFactorScheduler(step=step_, factor=factor) if len(step_) else None
def main():
if args.data_type == "cifar10":
args.aug_level = 1
args.num_classes = 10
# depth should be one of 110, 164, 1001,...,which is should fit (args.depth-2)%9 == 0
if((args.depth-2)%9 == 0 and args.depth >= 164):
per_unit = [(args.depth-2)/9]
filter_list = [16, 64, 128, 256]
bottle_neck = True
elif((args.depth-2)%6 == 0 and args.depth < 164):
per_unit = [(args.depth-2)/6]
filter_list = [16, 16, 32, 64]
bottle_neck = False
else:
raise ValueError("no experiments done on detph {}, you can do it youself".format(args.depth))
units = per_unit*3
symbol = resnext(units=units, num_stage=3, filter_list=filter_list, num_class=args.num_classes, num_group=args.num_group,
data_type="cifar10", drop_out=args.drop_out, ibn= True, bottle_neck = bottle_neck, bn_mom=args.bn_mom, workspace=args.workspace,
memonger=args.memonger)
elif args.data_type == "imagenet":
args.num_classes = 1000
if args.depth == 18:
units = [2, 2, 2, 2]
elif args.depth == 34:
units = [3, 4, 6, 3]
elif args.depth == 50:
units = [3, 4, 6, 3]
elif args.depth == 101:
units = [3, 4, 23, 3]
elif args.depth == 152:
units = [3, 8, 36, 3]
elif args.depth == 200:
units = [3, 24, 36, 3]
elif args.depth == 269:
units = [3, 30, 48, 8]
else:
raise ValueError("no experiments done on detph {}, you can do it youself".format(args.depth))
symbol = resnext(units=units, num_stage=4, filter_list=[64, 256, 512, 1024, 2048] if args.depth >=50
else [64, 64, 128, 256, 512], num_class=args.num_classes, num_group=args.num_group, data_type="imagenet", drop_out=args.drop_out, ibn= True, bottle_neck = True
if args.depth >= 50 else False, bn_mom=args.bn_mom, workspace=args.workspace,
memonger=args.memonger)
elif args.data_type == "vggface":
args.num_classes = 2613
if args.depth == 18:
units = [2, 2, 2, 2]
elif args.depth == 34:
units = [3, 4, 6, 3]
elif args.depth == 50:
units = [3, 4, 6, 3]
elif args.depth == 101:
units = [3, 4, 23, 3]
elif args.depth == 152:
units = [3, 8, 36, 3]
elif args.depth == 200:
units = [3, 24, 36, 3]
elif args.depth == 269:
units = [3, 30, 48, 8]
else:
raise ValueError("no experiments done on detph {}, you can do it youself".format(args.depth))
symbol = resnext(units=units, num_stage=4, filter_list=[64, 256, 512, 1024, 2048] if args.depth >=50
else [64, 64, 128, 256, 512], num_class=args.num_classes, num_group=args.num_group, data_type="vggface", drop_out=args.drop_out, ibn= True, bottle_neck = True
if args.depth >= 50 else False, bn_mom=args.bn_mom, workspace=args.workspace,
memonger=args.memonger)
elif args.data_type == "msface":
args.num_classes = 47003
if args.depth == 18:
units = [2, 2, 2, 2]
elif args.depth == 34:
units = [3, 4, 6, 3]
elif args.depth == 50:
units = [3, 4, 6, 3]
elif args.depth == 101:
units = [3, 4, 23, 3]
elif args.depth == 152:
units = [3, 8, 36, 3]
elif args.depth == 200:
units = [3, 24, 36, 3]
elif args.depth == 269:
units = [3, 30, 48, 8]
else:
raise ValueError("no experiments done on detph {}, you can do it youself".format(args.depth))
symbol = resnext(units=units, num_stage=4, filter_list=[64, 256, 512, 1024, 2048] if args.depth >=50
else [64, 64, 128, 256, 512], num_class=args.num_classes, num_group=args.num_group, data_type="msface", drop_out=args.drop_out, ibn= True, bottle_neck = True
if args.depth >= 50 else False, bn_mom=args.bn_mom, workspace=args.workspace,
memonger=args.memonger)
else:
raise ValueError("do not support {} yet".format(args.data_type))
kv = mx.kvstore.create(args.kv_store)
devs = mx.cpu() if args.gpus is None else [mx.gpu(int(i)) for i in args.gpus.split(',')]
epoch_size = max(int(args.num_examples / args.batch_size / kv.num_workers), 1)
begin_epoch = args.model_load_epoch if args.model_load_epoch else 0
if not os.path.exists("./model"):
os.mkdir("./model")
model_prefix = "model/resnext-ibn-a-wd-v2-{}-{}-{}".format(args.data_type, args.depth, kv.rank)
checkpoint = mx.callback.do_checkpoint(model_prefix)
arg_params = None
aux_params = None
if args.retrain:
_, arg_params, aux_params = mx.model.load_checkpoint(model_prefix, args.model_load_epoch)
if args.memonger:
import memonger
symbol = memonger.search_plan(symbol, data=(args.batch_size, 3, 32, 32) if args.data_type=="cifar10"
else (args.batch_size, 3, 224, 224))
train = mx.io.ImageRecordIter(
path_imgrec = os.path.join(args.data_dir, "train.rec") if args.data_type == 'cifar10' else
os.path.join(args.data_dir, "train_256_q90.rec") if args.aug_level == 1
else os.path.join(args.data_dir, "train_480_q90.rec") ,
label_width = 1,
data_name = 'data',
label_name = 'softmax_label',
data_shape = (3, 32, 32) if args.data_type=="cifar10" else (3, 224, 224),
batch_size = args.batch_size,
pad = 4 if args.data_type == "cifar10" else 0,
fill_value = 127, # only used when pad is valid
rand_crop = True,
max_random_scale = 1.0, # 480 with imagnet, 32 with cifar10
min_random_scale = 1.0 if args.data_type == "cifar10" else 1.0 if args.aug_level == 1 else 0.533, # 256.0/480.0=0.533, 256.0/384.0=0.667 256.0/256=1.0
max_aspect_ratio = 0 if args.data_type == "cifar10" else 0 if args.aug_level == 1 else 0.25, # 0.25
random_h = 0 if args.data_type == "cifar10" else 0 if args.aug_level == 1 else 36, # 0.4*90
random_s = 0 if args.data_type == "cifar10" else 0 if args.aug_level == 1 else 50, # 0.4*127
random_l = 0 if args.data_type == "cifar10" else 0 if args.aug_level == 1 else 50, # 0.4*127
max_rotate_angle = 0 if args.aug_level <= 2 else 10,
max_shear_ratio = 0 if args.aug_level <= 2 else 0.0, #0.1 args.aug_level = 3
rand_mirror = True,
shuffle = True,
num_parts = kv.num_workers,
part_index = kv.rank)
val = mx.io.ImageRecordIter(
path_imgrec = os.path.join(args.data_dir, "val.rec") if args.data_type == 'cifar10' else
os.path.join(args.data_dir, "val_256_q90.rec"),
label_width = 1,
data_name = 'data',
label_name = 'softmax_label',
batch_size = args.batch_size,
data_shape = (3, 32, 32) if args.data_type=="cifar10" else (3, 224, 224),
rand_crop = False,
rand_mirror = False,
num_parts = kv.num_workers,
part_index = kv.rank)
model = mx.model.FeedForward(
ctx = devs,
symbol = symbol,
arg_params = arg_params,
aux_params = aux_params,
num_epoch = 200 if args.data_type == "cifar10" else 125,
begin_epoch = begin_epoch,
learning_rate = args.lr,
momentum = args.mom,
wd = args.wd,
optimizer = 'nag',
# optimizer = 'sgd',
initializer = mx.init.Xavier(rnd_type='gaussian', factor_type="in", magnitude=2),
lr_scheduler = multi_factor_scheduler(begin_epoch, epoch_size, step=[220, 260, 280], factor=0.1)
if args.data_type=='cifar10' else
multi_factor_scheduler(begin_epoch, epoch_size, step=[30, 60, 90, 95, 115, 120], factor=0.1),
)
model.fit(
X = train,
eval_data = val,
eval_metric = ['acc'] if args.data_type=='cifar10' else
['acc', mx.metric.create('top_k_accuracy', top_k = 5), mx.metric.create('ce')],
kvstore = kv,
batch_end_callback = mx.callback.Speedometer(args.batch_size, args.frequent),
epoch_end_callback = checkpoint)
# logging.info("top-1 and top-5 acc is {}".format(model.score(X = val,
# eval_metric = ['acc', mx.metric.create('top_k_accuracy', top_k = 5)])))
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="command for training resnet-v2")
parser.add_argument('--gpus', type=str, default='0', help='the gpus will be used, e.g "0,1,2,3"')
parser.add_argument('--data-dir', type=str, default='./data/imagenet/', help='the input data directory')
parser.add_argument('--data-type', type=str, default='imagenet', help='the dataset type')
parser.add_argument('--list-dir', type=str, default='./',
help='the directory which contain the training list file')
parser.add_argument('--lr', type=float, default=0.1, help='initialization learning reate')
parser.add_argument('--mom', type=float, default=0.9, help='momentum for sgd')
parser.add_argument('--bn-mom', type=float, default=0.9, help='momentum for batch normlization')
parser.add_argument('--wd', type=float, default=0.0001, help='weight decay for sgd')
parser.add_argument('--batch-size', type=int, default=256, help='the batch size')
parser.add_argument('--num-group', type=int, default=32, help='the number of convolution groups')
parser.add_argument('--drop-out', type=float, default=0.0, help='the probability of an element to be zeroed')
parser.add_argument('--workspace', type=int, default=512, help='memory space size(MB) used in convolution, if xpu '
' memory is oom, then you can try smaller vale, such as --workspace 256')
parser.add_argument('--depth', type=int, default=50, help='the depth of resnet')
parser.add_argument('--num-classes', type=int, default=1000, help='the class number of your task')
parser.add_argument('--aug-level', type=int, default=2, choices=[1, 2, 3],
help='level 1: use only random crop and random mirror\n'
'level 2: add scale/aspect/hsv augmentation based on level 1\n'
'level 3: add rotation/shear augmentation based on level 2')
parser.add_argument('--num-examples', type=int, default=1281167, help='the number of training examples')
parser.add_argument('--kv-store', type=str, default='device', help='the kvstore type')
parser.add_argument('--model-load-epoch', type=int, default=0,
help='load the model on an epoch using the model-load-prefix')
parser.add_argument('--frequent', type=int, default=50, help='frequency of logging')
parser.add_argument('--memonger', action='store_true', default=False,
help='true means using memonger to save momory, https://github.com/dmlc/mxnet-memonger')
parser.add_argument('--retrain', action='store_true', default=False, help='true means continue training')
args = parser.parse_args()
hdlr = logging.FileHandler('./log/log-resnext-ibn-a-wd-v2-{}-{}-{}.log'.format(args.data_type, args.depth, 0))
hdlr.setFormatter(formatter)
logger.addHandler(hdlr)
logging.info(args)
main()