Skip to content

P
PaddleClas

PaddlePaddle / PaddleClas
大约 1 年 前同步成功

通知 115
Star 4999
Fork 1114
P
PaddleClas
提交 de4ebabf 编写于 作者: littletomatodonkey's avatar littletomatodonkey
浏览文件
操作

remove unused files

上级 03bbd585
显示空白变更内容
内联 并排
Showing with 0 addition and 1051 deletion
ppcls/modeling/architectures/csp_resnet.py 已删除 100644 → 0
浏览文件 @ 03bbd585
# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import math
import paddle.fluid as fluid
from paddle.fluid.param_attr import ParamAttr
__all__ = [
"CSPResNet50_leaky", "CSPResNet50_mish", "CSPResNet101_leaky",
"CSPResNet101_mish"
]
class CSPResNet():
def __init__(self, layers=50, act="leaky_relu"):
self.layers = layers
self.act = act
def net(self, input, class_dim=1000, data_format="NCHW"):
layers = self.layers
supported_layers = [50, 101]
assert layers in supported_layers, \
"supported layers are {} but input layer is {}".format(
supported_layers, layers)
if layers == 50:
depth = [3, 3, 5, 2]
elif layers == 101:
depth = [3, 3, 22, 2]
num_filters = [64, 128, 256, 512]
conv = self.conv_bn_layer(
input=input,
num_filters=64,
filter_size=7,
stride=2,
act=self.act,
name="conv1",
data_format=data_format)
conv = fluid.layers.pool2d(
input=conv,
pool_size=2,
pool_stride=2,
pool_padding=0,
pool_type='max',
data_format=data_format)
for block in range(len(depth)):
conv_name = "res" + str(block + 2) + chr(97)
if block != 0:
conv = self.conv_bn_layer(
input=conv,
num_filters=num_filters[block],
filter_size=3,
stride=2,
act=self.act,
name=conv_name + "_downsample",
data_format=data_format)
# split
left = conv
right = conv
if block == 0:
ch = num_filters[block]
else:
ch = num_filters[block] * 2
right = self.conv_bn_layer(
input=right,
num_filters=ch,
filter_size=1,
act=self.act,
name=conv_name + "_right_first_route",
data_format=data_format)
for i in range(depth[block]):
conv_name = "res" + str(block + 2) + chr(97 + i)
right = self.bottleneck_block(
input=right,
num_filters=num_filters[block],
stride=1,
name=conv_name,
data_format=data_format)
# route
left = self.conv_bn_layer(
input=left,
num_filters=num_filters[block] * 2,
filter_size=1,
act=self.act,
name=conv_name + "_left_route",
data_format=data_format)
right = self.conv_bn_layer(
input=right,
num_filters=num_filters[block] * 2,
filter_size=1,
act=self.act,
name=conv_name + "_right_route",
data_format=data_format)
conv = fluid.layers.concat([left, right], axis=1)
conv = self.conv_bn_layer(
input=conv,
num_filters=num_filters[block] * 2,
filter_size=1,
stride=1,
act=self.act,
name=conv_name + "_merged_transition",
data_format=data_format)
pool = fluid.layers.pool2d(
input=conv,
pool_type='avg',
global_pooling=True,
data_format=data_format)
stdv = 1.0 / math.sqrt(pool.shape[1] * 1.0)
out = fluid.layers.fc(
input=pool,
size=class_dim,
param_attr=fluid.param_attr.ParamAttr(
name="fc_0.w_0",
initializer=fluid.initializer.Uniform(-stdv, stdv)),
bias_attr=ParamAttr(name="fc_0.b_0"))
return out
def conv_bn_layer(self,
input,
num_filters,
filter_size,
stride=1,
groups=1,
act=None,
name=None,
data_format='NCHW'):
conv = fluid.layers.conv2d(
input=input,
num_filters=num_filters,
filter_size=filter_size,
stride=stride,
padding=(filter_size - 1) // 2,
groups=groups,
act=None,
param_attr=ParamAttr(name=name + "_weights"),
bias_attr=False,
name=name + '.conv2d.output.1',
data_format=data_format)
if name == "conv1":
bn_name = "bn_" + name
else:
bn_name = "bn" + name[3:]
bn = fluid.layers.batch_norm(
input=conv,
act=None,
name=bn_name + '.output.1',
param_attr=ParamAttr(name=bn_name + '_scale'),
bias_attr=ParamAttr(bn_name + '_offset'),
moving_mean_name=bn_name + '_mean',
moving_variance_name=bn_name + '_variance',
data_layout=data_format)
if act == "relu":
bn = fluid.layers.relu(bn)
elif act == "leaky_relu":
bn = fluid.layers.leaky_relu(bn)
elif act == "mish":
bn = self._mish(bn)
return bn
def _mish(self, input):
return input * fluid.layers.tanh(self._softplus(input))
def _softplus(self, input):
expf = fluid.layers.exp(fluid.layers.clip(input, -200, 50))
return fluid.layers.log(1 + expf)
def shortcut(self, input, ch_out, stride, is_first, name, data_format):
if data_format == 'NCHW':
ch_in = input.shape[1]
else:
ch_in = input.shape[-1]
if ch_in != ch_out or stride != 1 or is_first is True:
return self.conv_bn_layer(
input, ch_out, 1, stride, name=name, data_format=data_format)
else:
return input
def bottleneck_block(self, input, num_filters, stride, name, data_format):
conv0 = self.conv_bn_layer(
input=input,
num_filters=num_filters,
filter_size=1,
act="leaky_relu",
name=name + "_branch2a",
data_format=data_format)
conv1 = self.conv_bn_layer(
input=conv0,
num_filters=num_filters,
filter_size=3,
stride=stride,
act="leaky_relu",
name=name + "_branch2b",
data_format=data_format)
conv2 = self.conv_bn_layer(
input=conv1,
num_filters=num_filters * 2,
filter_size=1,
act=None,
name=name + "_branch2c",
data_format=data_format)
short = self.shortcut(
input,
num_filters * 2,
stride,
is_first=False,
name=name + "_branch1",
data_format=data_format)
ret = short + conv2
ret = fluid.layers.leaky_relu(ret, alpha=0.1)
return ret
def CSPResNet50_leaky():
model = CSPResNet(layers=50, act="leaky_relu")
return model
def CSPResNet50_mish():
model = CSPResNet(layers=50, act="mish")
return model
def CSPResNet101_leaky():
model = CSPResNet(layers=101, act="leaky_relu")
return model
def CSPResNet101_mish():
model = CSPResNet(layers=101, act="mish")
return model
ppcls/modeling/architectures/darts_gs.py 已删除 100644 → 0
浏览文件 @ 03bbd585
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
#
#Licensed under the Apache License, Version 2.0 (the "License");
#you may not use this file except in compliance with the License.
#You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
#Unless required by applicable law or agreed to in writing, software
#distributed under the License is distributed on an "AS IS" BASIS,
#WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
#See the License for the specific language governing permissions and
#limitations under the License.
#
# Based on:
# --------------------------------------------------------
# DARTS
# Copyright (c) 2018, Hanxiao Liu.
# Licensed under the Apache License, Version 2.0;
# --------------------------------------------------------
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import os
import sys
import numpy as np
import time
import functools
import paddle
import paddle.fluid as fluid
from paddle.fluid.param_attr import ParamAttr
from paddle.fluid.initializer import Xavier
from paddle.fluid.initializer import Normal
from paddle.fluid.initializer import Constant
from collections import namedtuple
Genotype = namedtuple('Genotype', 'normal normal_concat reduce reduce_concat')
arch_dict = {
'DARTS_GS_6M': Genotype(
normal=[('sep_conv_3x3', 1), ('sep_conv_3x3', 0), ('sep_conv_5x5', 1),
('sep_conv_5x5', 0), ('sep_conv_3x3', 2), ('sep_conv_3x3', 1),
('skip_connect', 4), ('sep_conv_3x3', 3)],
normal_concat=range(2, 6),
reduce=[('sep_conv_5x5', 0), ('max_pool_3x3', 1), ('dil_conv_5x5', 2),
('sep_conv_5x5', 0), ('sep_conv_3x3', 1), ('dil_conv_5x5', 3),
('dil_conv_3x3', 1), ('sep_conv_3x3', 2)],
reduce_concat=range(2, 6)),
'DARTS_GS_4M': Genotype(
normal=[('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('sep_conv_3x3', 0),
('sep_conv_3x3', 1), ('sep_conv_3x3', 1), ('skip_connect', 0),
('skip_connect', 0), ('dil_conv_3x3', 1)],
normal_concat=range(2, 6),
reduce=[('max_pool_3x3', 0), ('max_pool_3x3', 1), ('max_pool_3x3', 0),
('avg_pool_3x3', 1), ('skip_connect', 3), ('skip_connect', 2),
('sep_conv_3x3', 0), ('sep_conv_5x5', 2)],
reduce_concat=range(2, 6)),
}
__all__ = list(arch_dict.keys())
OPS = {
'none' : lambda input, C, stride, name, affine: Zero(input, stride, name),
'avg_pool_3x3' : lambda input, C, stride, name, affine: fluid.layers.pool2d(input, 3, 'avg', pool_stride=stride, pool_padding=1, name=name),
'max_pool_3x3' : lambda input, C, stride, name, affine: fluid.layers.pool2d(input, 3, 'max', pool_stride=stride, pool_padding=1, name=name),
'skip_connect' : lambda input,C, stride, name, affine: Identity(input, name) if stride == 1 else FactorizedReduce(input, C, name=name, affine=affine),
'sep_conv_3x3' : lambda input,C, stride, name, affine: SepConv(input, C, C, 3, stride, 1, name=name, affine=affine),
'sep_conv_5x5' : lambda input,C, stride, name, affine: SepConv(input, C, C, 5, stride, 2, name=name, affine=affine),
'sep_conv_7x7' : lambda input,C, stride, name, affine: SepConv(input, C, C, 7, stride, 3, name=name, affine=affine),
'dil_conv_3x3' : lambda input,C, stride, name, affine: DilConv(input, C, C, 3, stride, 2, 2, name=name, affine=affine),
'dil_conv_5x5' : lambda input,C, stride, name, affine: DilConv(input, C, C, 5, stride, 4, 2, name=name, affine=affine),
'conv_7x1_1x7' : lambda input,C, stride, name, affine: SevenConv(input, C, name=name, affine=affine)
}
def ReLUConvBN(input,
C_out,
kernel_size,
stride,
padding,
name='',
affine=True):
relu_a = fluid.layers.relu(input)
conv2d_a = fluid.layers.conv2d(
relu_a, C_out, kernel_size, stride, padding, bias_attr=False)
if affine:
reluconvbn_out = fluid.layers.batch_norm(
conv2d_a,
param_attr=ParamAttr(
initializer=Constant(1.), name=name + 'op.2.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.), name=name + 'op.2.bias'),
moving_mean_name=name + 'op.2.running_mean',
moving_variance_name=name + 'op.2.running_var')
else:
reluconvbn_out = fluid.layers.batch_norm(
conv2d_a,
param_attr=ParamAttr(
initializer=Constant(1.),
learning_rate=0.,
name=name + 'op.2.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.),
learning_rate=0.,
name=name + 'op.2.bias'),
moving_mean_name=name + 'op.2.running_mean',
moving_variance_name=name + 'op.2.running_var')
return reluconvbn_out
def DilConv(input,
C_in,
C_out,
kernel_size,
stride,
padding,
dilation,
name='',
affine=True):
relu_a = fluid.layers.relu(input)
conv2d_a = fluid.layers.conv2d(
relu_a,
C_in,
kernel_size,
stride,
padding,
dilation,
groups=C_in,
bias_attr=False,
use_cudnn=False)
conv2d_b = fluid.layers.conv2d(conv2d_a, C_out, 1, bias_attr=False)
if affine:
dilconv_out = fluid.layers.batch_norm(
conv2d_b,
param_attr=ParamAttr(
initializer=Constant(1.), name=name + 'op.3.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.), name=name + 'op.3.bias'),
moving_mean_name=name + 'op.3.running_mean',
moving_variance_name=name + 'op.3.running_var')
else:
dilconv_out = fluid.layers.batch_norm(
conv2d_b,
param_attr=ParamAttr(
initializer=Constant(1.),
learning_rate=0.,
name=name + 'op.3.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.),
learning_rate=0.,
name=name + 'op.3.bias'),
moving_mean_name=name + 'op.3.running_mean',
moving_variance_name=name + 'op.3.running_var')
return dilconv_out
def SepConv(input,
C_in,
C_out,
kernel_size,
stride,
padding,
name='',
affine=True):
relu_a = fluid.layers.relu(input)
conv2d_a = fluid.layers.conv2d(
relu_a,
C_in,
kernel_size,
stride,
padding,
groups=C_in,
bias_attr=False,
use_cudnn=False)
conv2d_b = fluid.layers.conv2d(conv2d_a, C_in, 1, bias_attr=False)
if affine:
bn_a = fluid.layers.batch_norm(
conv2d_b,
param_attr=ParamAttr(
initializer=Constant(1.), name=name + 'op.3.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.), name=name + 'op.3.bias'),
moving_mean_name=name + 'op.3.running_mean',
moving_variance_name=name + 'op.3.running_var')
else:
bn_a = fluid.layers.batch_norm(
conv2d_b,
param_attr=ParamAttr(
initializer=Constant(1.),
learning_rate=0.,
name=name + 'op.3.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.),
learning_rate=0.,
name=name + 'op.3.bias'),
moving_mean_name=name + 'op.3.running_mean',
moving_variance_name=name + 'op.3.running_var')
relu_b = fluid.layers.relu(bn_a)
conv2d_d = fluid.layers.conv2d(
relu_b,
C_in,
kernel_size,
1,
padding,
groups=C_in,
bias_attr=False,
use_cudnn=False)
conv2d_e = fluid.layers.conv2d(conv2d_d, C_out, 1, bias_attr=False)
if affine:
sepconv_out = fluid.layers.batch_norm(
conv2d_e,
param_attr=ParamAttr(
initializer=Constant(1.), name=name + 'op.7.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.), name=name + 'op.7.bias'),
moving_mean_name=name + 'op.7.running_mean',
moving_variance_name=name + 'op.7.running_var')
else:
sepconv_out = fluid.layers.batch_norm(
conv2d_e,
param_attr=ParamAttr(
initializer=Constant(1.),
learning_rate=0.,
name=name + 'op.7.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.),
learning_rate=0.,
name=name + 'op.7.bias'),
moving_mean_name=name + 'op.7.running_mean',
moving_variance_name=name + 'op.7.running_var')
return sepconv_out
def SevenConv(input, C_out, stride, name='', affine=True):
relu_a = fluid.layers.relu(input)
conv2d_a = fluid.layers.conv2d(
relu_a,
C_out, (1, 7), (1, stride), (0, 3),
param_attr=ParamAttr(
initializer=Xavier(
uniform=False, fan_in=0),
name=name + 'op.1.weight'),
bias_attr=False)
conv2d_b = fluid.layers.conv2d(
conv2d_a,
C_out, (7, 1), (stride, 1), (3, 0),
param_attr=ParamAttr(
initializer=Xavier(
uniform=False, fan_in=0),
name=name + 'op.2.weight'),
bias_attr=False)
if affine:
out = fluid.layers.batch_norm(
conv2d_b,
param_attr=ParamAttr(
initializer=Constant(1.), name=name + 'op.3.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.), name=name + 'op.3.bias'),
moving_mean_name=name + 'op.3.running_mean',
moving_variance_name=name + 'op.3.running_var')
else:
out = fluid.layers.batch_norm(
conv2d_b,
param_attr=ParamAttr(
initializer=Constant(1.),
learning_rate=0.,
name=name + 'op.3.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.),
learning_rate=0.,
name=name + 'op.3.bias'),
moving_mean_name=name + 'op.3.running_mean',
moving_variance_name=name + 'op.3.running_var')
def Identity(input, name=''):
return input
def Zero(input, stride, name=''):
ones = np.ones(input.shape[-2:])
ones[::stride, ::stride] = 0
ones = fluid.layers.assign(ones)
return input * ones
def FactorizedReduce(input, C_out, name='', affine=True):
relu_a = fluid.layers.relu(input)
conv2d_a = fluid.layers.conv2d(
relu_a,
C_out // 2,
1,
2,
param_attr=ParamAttr(
initializer=Xavier(
uniform=False, fan_in=0),
name=name + 'conv_1.weight'),
bias_attr=False)
h_end = relu_a.shape[2]
w_end = relu_a.shape[3]
slice_a = fluid.layers.slice(relu_a, [2, 3], [1, 1], [h_end, w_end])
conv2d_b = fluid.layers.conv2d(
slice_a,
C_out // 2,
1,
2,
param_attr=ParamAttr(
initializer=Xavier(
uniform=False, fan_in=0),
name=name + 'conv_2.weight'),
bias_attr=False)
out = fluid.layers.concat([conv2d_a, conv2d_b], axis=1)
if affine:
out = fluid.layers.batch_norm(
out,
param_attr=ParamAttr(
initializer=Constant(1.), name=name + 'bn.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.), name=name + 'bn.bias'),
moving_mean_name=name + 'bn.running_mean',
moving_variance_name=name + 'bn.running_var')
else:
out = fluid.layers.batch_norm(
out,
param_attr=ParamAttr(
initializer=Constant(1.),
learning_rate=0.,
name=name + 'bn.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.),
learning_rate=0.,
name=name + 'bn.bias'),
moving_mean_name=name + 'bn.running_mean',
moving_variance_name=name + 'bn.running_var')
return out
class Cell():
def __init__(self, genotype, C_prev_prev, C_prev, C, reduction,
reduction_prev):
if reduction_prev:
self.preprocess0 = functools.partial(FactorizedReduce, C_out=C)
else:
self.preprocess0 = functools.partial(
ReLUConvBN, C_out=C, kernel_size=1, stride=1, padding=0)
self.preprocess1 = functools.partial(
ReLUConvBN, C_out=C, kernel_size=1, stride=1, padding=0)
if reduction:
op_names, indices = zip(*genotype.reduce)
concat = genotype.reduce_concat
else:
op_names, indices = zip(*genotype.normal)
concat = genotype.normal_concat
print(op_names, indices, concat, reduction)
self._compile(C, op_names, indices, concat, reduction)
def _compile(self, C, op_names, indices, concat, reduction):
assert len(op_names) == len(indices)
self._steps = len(op_names) // 2
self._concat = concat
self.multiplier = len(concat)
self._ops = []
for name, index in zip(op_names, indices):
stride = 2 if reduction and index < 2 else 1
op = functools.partial(OPS[name], C=C, stride=stride, affine=True)
self._ops += [op]
self._indices = indices
def forward(self, s0, s1, drop_prob, is_train, name):
self.training = is_train
preprocess0_name = name + 'preprocess0.'
preprocess1_name = name + 'preprocess1.'
s0 = self.preprocess0(s0, name=preprocess0_name)
s1 = self.preprocess1(s1, name=preprocess1_name)
out = [s0, s1]
for i in range(self._steps):
h1 = out[self._indices[2 * i]]
h2 = out[self._indices[2 * i + 1]]
op1 = self._ops[2 * i]
op2 = self._ops[2 * i + 1]
h3 = op1(h1, name=name + '_ops.' + str(2 * i) + '.')
h4 = op2(h2, name=name + '_ops.' + str(2 * i + 1) + '.')
if self.training and drop_prob > 0.:
if h3 != h1:
h3 = fluid.layers.dropout(
h3,
drop_prob,
dropout_implementation='upscale_in_train')
if h4 != h2:
h4 = fluid.layers.dropout(
h4,
drop_prob,
dropout_implementation='upscale_in_train')
s = h3 + h4
out += [s]
return fluid.layers.concat([out[i] for i in self._concat], axis=1)
def AuxiliaryHeadImageNet(input, num_classes, aux_name='auxiliary_head'):
relu_a = fluid.layers.relu(input)
pool_a = fluid.layers.pool2d(relu_a, 5, 'avg', 2)
conv2d_a = fluid.layers.conv2d(
pool_a, 128, 1, name=aux_name + '.features.2', bias_attr=False)
bn_a_name = aux_name + '.features.3'
bn_a = fluid.layers.batch_norm(
conv2d_a,
act='relu',
name=bn_a_name,
param_attr=ParamAttr(
initializer=Constant(1.), name=bn_a_name + '.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.), name=bn_a_name + '.bias'),
moving_mean_name=bn_a_name + '.running_mean',
moving_variance_name=bn_a_name + '.running_var')
conv2d_b = fluid.layers.conv2d(
bn_a, 768, 2, name=aux_name + '.features.5', bias_attr=False)
bn_b_name = aux_name + '.features.6'
bn_b = fluid.layers.batch_norm(
conv2d_b,
act='relu',
name=bn_b_name,
param_attr=ParamAttr(
initializer=Constant(1.), name=bn_b_name + '.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.), name=bn_b_name + '.bias'),
moving_mean_name=bn_b_name + '.running_mean',
moving_variance_name=bn_b_name + '.running_var')
pool_b = fluid.layers.adaptive_pool2d(bn_b, (1, 1), "avg")
fc_name = aux_name + '.classifier'
fc = fluid.layers.fc(pool_b,
num_classes,
name=fc_name,
param_attr=ParamAttr(
initializer=Normal(scale=1e-3),
name=fc_name + '.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.), name=fc_name + '.bias'))
return fc
def StemConv0(input, C_out):
conv_a = fluid.layers.conv2d(
input, C_out // 2, 3, stride=2, padding=1, bias_attr=False)
bn_a = fluid.layers.batch_norm(
conv_a,
act='relu',
param_attr=ParamAttr(
initializer=Constant(1.), name='stem0.1.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.), name='stem0.1.bias'),
moving_mean_name='stem0.1.running_mean',
moving_variance_name='stem0.1.running_var')
conv_b = fluid.layers.conv2d(
bn_a, C_out, 3, stride=2, padding=1, bias_attr=False)
bn_b = fluid.layers.batch_norm(
conv_b,
param_attr=ParamAttr(
initializer=Constant(1.), name='stem0.3.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.), name='stem0.3.bias'),
moving_mean_name='stem0.3.running_mean',
moving_variance_name='stem0.3.running_var')
return bn_b
def StemConv1(input, C_out):
relu_a = fluid.layers.relu(input)
conv_a = fluid.layers.conv2d(
relu_a, C_out, 3, stride=2, padding=1, bias_attr=False)
bn_a = fluid.layers.batch_norm(
conv_a,
param_attr=ParamAttr(
initializer=Constant(1.), name='stem1.1.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.), name='stem1.1.bias'),
moving_mean_name='stem1.1.running_mean',
moving_variance_name='stem1.1.running_var')
return bn_a
class NetworkImageNet(object):
def __init__(self, arch='DARTS_6M'):
self.class_num = 1000
self.init_channel = 48
self._layers = 14
self._auxiliary = False
self.drop_path_prob = 0
genotype = arch_dict[arch]
C = self.init_channel
layers = self._layers
C_prev_prev, C_prev, C_curr = C, C, C
self.cells = []
reduction_prev = True
for i in range(layers):
if i in [layers // 3, 2 * layers // 3]:
C_curr *= 2
reduction = True
else:
reduction = False
cell = Cell(genotype, C_prev_prev, C_prev, C_curr, reduction,
reduction_prev)
reduction_prev = reduction
self.cells += [cell]
C_prev_prev, C_prev = C_prev, cell.multiplier * C_curr
if i == 2 * layers // 3:
C_to_auxiliary = C_prev
def net(self, input, class_dim=1000, is_train=True):
self.logits_aux = None
num_channel = self.init_channel
s0 = StemConv0(input, num_channel)
s1 = StemConv1(s0, num_channel)
for i, cell in enumerate(self.cells):
name = 'cells.' + str(i) + '.'
s0, s1 = s1, cell.forward(s0, s1, self.drop_path_prob, is_train,
name)
if i == int(2 * self._layers // 3):
if self._auxiliary and is_train:
self.logits_aux = AuxiliaryHeadImageNet(s1, self.class_num)
out = fluid.layers.adaptive_pool2d(s1, (1, 1), "avg")
self.logits = fluid.layers.fc(out,
size=self.class_num,
param_attr=ParamAttr(
initializer=Normal(scale=1e-4),
name='classifier.weight'),
bias_attr=ParamAttr(
initializer=Constant(0.),
name='classifier.bias'))
return self.logits
def DARTS_GS_6M():
return NetworkImageNet(arch='DARTS_GS_6M')
def DARTS_GS_4M():
return NetworkImageNet(arch='DARTS_GS_4M')
ppcls/modeling/architectures/layers.py 已删除 100644 → 0
浏览文件 @ 03bbd585
#copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
#
#Licensed under the Apache License, Version 2.0 (the "License");
#you may not use this file except in compliance with the License.
#You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
#Unless required by applicable law or agreed to in writing, software
#distributed under the License is distributed on an "AS IS" BASIS,
#WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
#See the License for the specific language governing permissions and
#limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import math
import warnings
import paddle.fluid as fluid
def initial_type(name,
input,
op_type,
fan_out,
init="google",
use_bias=False,
filter_size=0,
stddev=0.02):
if init == "kaiming":
if op_type == 'conv':
fan_in = input.shape[1] * filter_size * filter_size
elif op_type == 'deconv':
fan_in = fan_out * filter_size * filter_size
else:
if len(input.shape) > 2:
fan_in = input.shape[1] * input.shape[2] * input.shape[3]
else:
fan_in = input.shape[1]
bound = 1 / math.sqrt(fan_in)
param_attr = fluid.ParamAttr(
name=name + "_weights",
initializer=fluid.initializer.Uniform(
low=-bound, high=bound))
if use_bias == True:
bias_attr = fluid.ParamAttr(
name=name + '_offset',
initializer=fluid.initializer.Uniform(
low=-bound, high=bound))
else:
bias_attr = False
elif init == 'google':
n = filter_size * filter_size * fan_out
param_attr = fluid.ParamAttr(
name=name + "_weights",
initializer=fluid.initializer.NormalInitializer(
loc=0.0, scale=math.sqrt(2.0 / n)))
if use_bias == True:
bias_attr = fluid.ParamAttr(
name=name + "_offset",
initializer=fluid.initializer.Constant(0.0))
else:
bias_attr = False
else:
param_attr = fluid.ParamAttr(
name=name + "_weights",
initializer=fluid.initializer.NormalInitializer(
loc=0.0, scale=stddev))
if use_bias == True:
bias_attr = fluid.ParamAttr(
name=name + "_offset",
initializer=fluid.initializer.Constant(0.0))
else:
bias_attr = False
return param_attr, bias_attr
def cal_padding(img_size, stride, filter_size, dilation=1):
"""Calculate padding size."""
if img_size % stride == 0:
out_size = max(filter_size - stride, 0)
else:
out_size = max(filter_size - (img_size % stride), 0)
return out_size // 2, out_size - out_size // 2
def init_batch_norm_layer(name="batch_norm"):
param_attr = fluid.ParamAttr(
name=name + '_scale', initializer=fluid.initializer.Constant(1.0))
bias_attr = fluid.ParamAttr(
name=name + '_offset',
initializer=fluid.initializer.Constant(value=0.0))
return param_attr, bias_attr
def init_fc_layer(fout, name='fc'):
n = fout # fan-out
init_range = 1.0 / math.sqrt(n)
param_attr = fluid.ParamAttr(
name=name + '_weights',
initializer=fluid.initializer.UniformInitializer(
low=-init_range, high=init_range))
bias_attr = fluid.ParamAttr(
name=name + '_offset',
initializer=fluid.initializer.Constant(value=0.0))
return param_attr, bias_attr
def norm_layer(input, norm_type='batch_norm', name=None):
if norm_type == 'batch_norm':
param_attr = fluid.ParamAttr(
name=name + '_weights',
initializer=fluid.initializer.Constant(1.0))
bias_attr = fluid.ParamAttr(
name=name + '_offset',
initializer=fluid.initializer.Constant(value=0.0))
return fluid.layers.batch_norm(
input,
param_attr=param_attr,
bias_attr=bias_attr,
moving_mean_name=name + '_mean',
moving_variance_name=name + '_variance')
elif norm_type == 'instance_norm':
helper = fluid.layer_helper.LayerHelper("instance_norm", **locals())
dtype = helper.input_dtype()
epsilon = 1e-5
mean = fluid.layers.reduce_mean(input, dim=[2, 3], keep_dim=True)
var = fluid.layers.reduce_mean(
fluid.layers.square(input - mean), dim=[2, 3], keep_dim=True)
if name is not None:
scale_name = name + "_scale"
offset_name = name + "_offset"
scale_param = fluid.ParamAttr(
name=scale_name,
initializer=fluid.initializer.Constant(1.0),
trainable=True)
offset_param = fluid.ParamAttr(
name=offset_name,
initializer=fluid.initializer.Constant(0.0),
trainable=True)
scale = helper.create_parameter(
attr=scale_param, shape=input.shape[1:2], dtype=dtype)
offset = helper.create_parameter(
attr=offset_param, shape=input.shape[1:2], dtype=dtype)
tmp = fluid.layers.elementwise_mul(x=(input - mean), y=scale, axis=1)
tmp = tmp / fluid.layers.sqrt(var + epsilon)
tmp = fluid.layers.elementwise_add(tmp, offset, axis=1)
return tmp
else:
raise NotImplementedError("norm tyoe: [%s] is not support" % norm_type)
def conv2d(input,
num_filters=64,
filter_size=7,
stride=1,
stddev=0.02,
padding=0,
groups=None,
name="conv2d",
norm=None,
act=None,
relufactor=0.0,
use_bias=False,
padding_type=None,
initial="normal",
use_cudnn=True):
if padding != 0 and padding_type != None:
warnings.warn(
'padding value and padding type are set in the same time, and the final padding width and padding height are computed by padding_type'
)
param_attr, bias_attr = initial_type(
name=name,
input=input,
op_type='conv',
fan_out=num_filters,
init=initial,
use_bias=use_bias,
filter_size=filter_size,
stddev=stddev)
def get_padding(filter_size, stride=1, dilation=1):
padding = ((stride - 1) + dilation * (filter_size - 1)) // 2
return padding
need_crop = False
if padding_type == "SAME":
top_padding, bottom_padding = cal_padding(input.shape[2], stride,
filter_size)
left_padding, right_padding = cal_padding(input.shape[2], stride,
filter_size)
height_padding = bottom_padding
width_padding = right_padding
if top_padding != bottom_padding or left_padding != right_padding:
height_padding = top_padding + stride
width_padding = left_padding + stride
need_crop = True
padding = [height_padding, width_padding]
elif padding_type == "VALID":
height_padding = 0
width_padding = 0
padding = [height_padding, width_padding]
elif padding_type == "DYNAMIC":
padding = get_padding(filter_size, stride)
else:
padding = padding
conv = fluid.layers.conv2d(
input,
num_filters,
filter_size,
groups=groups,
name=name,
stride=stride,
padding=padding,
use_cudnn=use_cudnn,
param_attr=param_attr,
bias_attr=bias_attr)
if need_crop:
conv = conv[:, :, 1:, 1:]
if norm is not None:
conv = norm_layer(input=conv, norm_type=norm, name=name + "_norm")
if act == 'relu':
conv = fluid.layers.relu(conv, name=name + '_relu')
elif act == 'leaky_relu':
conv = fluid.layers.leaky_relu(
conv, alpha=relufactor, name=name + '_leaky_relu')
elif act == 'tanh':
conv = fluid.layers.tanh(conv, name=name + '_tanh')
elif act == 'sigmoid':
conv = fluid.layers.sigmoid(conv, name=name + '_sigmoid')
elif act == 'swish':
conv = fluid.layers.swish(conv, name=name + '_swish')
elif act == None:
conv = conv
else:
raise NotImplementedError("activation: [%s] is not support" % act)
return conv
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或