Skip to content

GitLab

Unverified Commit 4ffb5b62 authored by zhoujun's avatar zhoujun Committed by GitHub
Browse files

Merge pull request #924 from WenmuZhou/dygraph 

Dygraph
parents bc93c549 aad3093a
Hide whitespace changes
Inline Side-by-side
Showing with 1489 additions and 0 deletions
ppocr/modeling/architectures/__init__.py 0 → 100755
View file @ 4ffb5b62
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# 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 .model import Model
__all__ = ['Model']
\ No newline at end of file
ppocr/modeling/architectures/model.py 0 → 100644
View file @ 4ffb5b62
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# 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 os, sys
__dir__ = os.path.dirname(os.path.abspath(__file__))
sys.path.append(__dir__)
sys.path.append('/home/zhoujun20/PaddleOCR')
import paddle
from paddle import nn
from ppocr.modeling.transform import build_transform
from ppocr.modeling.backbones import build_backbone
from ppocr.modeling.necks import build_neck
from ppocr.modeling.heads import build_head
__all__ = ['Model']
class Model(nn.Layer):
def __init__(self, config):
"""
Detection module for OCR.
args:
config (dict): the super parameters for module.
"""
super(Model, self).__init__()
algorithm = config['algorithm']
self.type = config['type']
self.model_name = '{}_{}'.format(self.type, algorithm)
in_channels = config.get('in_channels', 3)
# build transfrom,
# for rec, transfrom can be TPS,None
# for det and cls, transfrom shoule to be None,
# if you make model differently, you can use transfrom in det and cls
if 'Transform' not in config or config['Transform'] is None:
self.use_transform = False
else:
self.use_transform = True
config['Transform']['in_channels'] = in_channels
self.transform = build_transform(config['Transform'])
in_channels = self.transform.out_channels
# build backbone, backbone is need for del, rec and cls
config["Backbone"]['in_channels'] = in_channels
self.backbone = build_backbone(config["Backbone"], self.type)
in_channels = self.backbone.out_channels
# build neck
# for rec, neck can be cnn,rnn or reshape(None)
# for det, neck can be FPN, BIFPN and so on.
# for cls, neck should be none
if 'Neck' not in config or config['Neck'] is None:
self.use_neck = False
else:
self.use_neck = True
config['Neck']['in_channels'] = in_channels
self.neck = build_neck(config['Neck'])
in_channels = self.neck.out_channels
# # build head, head is need for del, rec and cls
config["Head"]['in_channels'] = in_channels
self.head = build_head(config["Head"])
# @paddle.jit.to_static
def forward(self, x):
if self.use_transform:
x = self.transform(x)
x = self.backbone(x)
if self.use_neck:
x = self.neck(x)
x = self.head(x)
return x
def check_static():
import numpy as np
from ppocr.utils.save_load import load_dygraph_pretrain
from ppocr.utils.logging import get_logger
from tools import program
config = program.load_config('configs/det/det_r50_vd_db.yml')
# import cv2
# data = cv2.imread('doc/imgs/1.jpg')
# data = normalize(data)
logger = get_logger()
data = np.zeros((1, 3, 640, 640), dtype=np.float32)
paddle.disable_static()
config['Architecture']['in_channels'] = 3
config['Architecture']["Head"]['out_channels'] = 6624
model = Model(config['Architecture'])
model.eval()
load_dygraph_pretrain(
model,
logger,
'/Users/zhoujun20/Desktop/code/PaddleOCR/db/db',
load_static_weights=True)
x = paddle.to_variable(data)
y = model(x)
for y1 in y:
print(y1.shape)
#
# # from matplotlib import pyplot as plt
# # plt.imshow(y.numpy())
# # plt.show()
static_out = np.load('/Users/zhoujun20/Desktop/code/PaddleOCR/db/db.npy')
diff = y.numpy() - static_out
print(y.shape, static_out.shape, diff.mean())
if __name__ == '__main__':
check_static()
ppocr/modeling/backbones/__init__.py 0 → 100755
View file @ 4ffb5b62
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# 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.
__all__ = ['build_backbone']
def build_backbone(config, model_type):
if model_type == 'det':
from .det_mobilenet_v3 import MobileNetV3
from .det_resnet_vd import ResNet
support_dict = ['MobileNetV3', 'ResNet', 'ResNet_SAST']
elif model_type == 'rec':
from .rec_mobilenet_v3 import MobileNetV3
from .rec_resnet_vd import ResNet
support_dict = ['MobileNetV3', 'ResNet', 'ResNet_FPN']
else:
raise NotImplementedError
module_name = config.pop('name')
assert module_name in support_dict, Exception(
'when model typs is {}, backbone only support {}'.format(model_type,
support_dict))
module_class = eval(module_name)(**config)
return module_class
ppocr/modeling/backbones/det_mobilenet_v3.py 0 → 100755
View file @ 4ffb5b62
# 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 paddle
from paddle import nn
import paddle.nn.functional as F
from paddle import ParamAttr
__all__ = ['MobileNetV3']
def make_divisible(v, divisor=8, min_value=None):
if min_value is None:
min_value = divisor
new_v = max(min_value, int(v + divisor / 2) // divisor * divisor)
if new_v < 0.9 * v:
new_v += divisor
return new_v
class MobileNetV3(nn.Layer):
def __init__(self, in_channels=3, model_name='large', scale=0.5, **kwargs):
"""
the MobilenetV3 backbone network for detection module.
Args:
params(dict): the super parameters for build network
"""
super(MobileNetV3, self).__init__()
if model_name == "large":
cfg = [
# k, exp, c, se, nl, s,
[3, 16, 16, False, 'relu', 1],
[3, 64, 24, False, 'relu', 2],
[3, 72, 24, False, 'relu', 1],
[5, 72, 40, True, 'relu', 2],
[5, 120, 40, True, 'relu', 1],
[5, 120, 40, True, 'relu', 1],
[3, 240, 80, False, 'hard_swish', 2],
[3, 200, 80, False, 'hard_swish', 1],
[3, 184, 80, False, 'hard_swish', 1],
[3, 184, 80, False, 'hard_swish', 1],
[3, 480, 112, True, 'hard_swish', 1],
[3, 672, 112, True, 'hard_swish', 1],
[5, 672, 160, True, 'hard_swish', 2],
[5, 960, 160, True, 'hard_swish', 1],
[5, 960, 160, True, 'hard_swish', 1],
]
cls_ch_squeeze = 960
elif model_name == "small":
cfg = [
# k, exp, c, se, nl, s,
[3, 16, 16, True, 'relu', 2],
[3, 72, 24, False, 'relu', 2],
[3, 88, 24, False, 'relu', 1],
[5, 96, 40, True, 'hard_swish', 2],
[5, 240, 40, True, 'hard_swish', 1],
[5, 240, 40, True, 'hard_swish', 1],
[5, 120, 48, True, 'hard_swish', 1],
[5, 144, 48, True, 'hard_swish', 1],
[5, 288, 96, True, 'hard_swish', 2],
[5, 576, 96, True, 'hard_swish', 1],
[5, 576, 96, True, 'hard_swish', 1],
]
cls_ch_squeeze = 576
else:
raise NotImplementedError("mode[" + model_name +
"_model] is not implemented!")
supported_scale = [0.35, 0.5, 0.75, 1.0, 1.25]
assert scale in supported_scale, \
"supported scale are {} but input scale is {}".format(supported_scale, scale)
inplanes = 16
# conv1
self.conv = ConvBNLayer(
in_channels=in_channels,
out_channels=make_divisible(inplanes * scale),
kernel_size=3,
stride=2,
padding=1,
groups=1,
if_act=True,
act='hard_swish',
name='conv1')
self.stages = []
self.out_channels = []
block_list = []
i = 0
inplanes = make_divisible(inplanes * scale)
for (k, exp, c, se, nl, s) in cfg:
if s == 2 and i > 2:
self.out_channels.append(inplanes)
self.stages.append(nn.Sequential(*block_list))
block_list = []
block_list.append(
ResidualUnit(
in_channels=inplanes,
mid_channels=make_divisible(scale * exp),
out_channels=make_divisible(scale * c),
kernel_size=k,
stride=s,
use_se=se,
act=nl,
name="conv" + str(i + 2)))
inplanes = make_divisible(scale * c)
i += 1
block_list.append(
ConvBNLayer(
in_channels=inplanes,
out_channels=make_divisible(scale * cls_ch_squeeze),
kernel_size=1,
stride=1,
padding=0,
groups=1,
if_act=True,
act='hard_swish',
name='conv_last'))
self.stages.append(nn.Sequential(*block_list))
self.out_channels.append(make_divisible(scale * cls_ch_squeeze))
for i, stage in enumerate(self.stages):
self.add_sublayer(sublayer=stage, name="stage{}".format(i))
def forward(self, x):
x = self.conv(x)
out_list = []
for stage in self.stages:
x = stage(x)
out_list.append(x)
return out_list
class ConvBNLayer(nn.Layer):
def __init__(self,
in_channels,
out_channels,
kernel_size,
stride,
padding,
groups=1,
if_act=True,
act=None,
name=None):
super(ConvBNLayer, self).__init__()
self.if_act = if_act
self.act = act
self.conv = nn.Conv2d(
in_channels=in_channels,
out_channels=out_channels,
kernel_size=kernel_size,
stride=stride,
padding=padding,
groups=groups,
weight_attr=ParamAttr(name=name + '_weights'),
bias_attr=False)
self.bn = nn.BatchNorm(
num_channels=out_channels,
act=None,
param_attr=ParamAttr(name=name + "_bn_scale"),
bias_attr=ParamAttr(name=name + "_bn_offset"),
moving_mean_name=name + "_bn_mean",
moving_variance_name=name + "_bn_variance")
def forward(self, x):
x = self.conv(x)
x = self.bn(x)
if self.if_act:
if self.act == "relu":
x = F.relu(x)
elif self.act == "hard_swish":
x = F.hard_swish(x)
else:
print("The activation function is selected incorrectly.")
exit()
return x
class ResidualUnit(nn.Layer):
def __init__(self,
in_channels,
mid_channels,
out_channels,
kernel_size,
stride,
use_se,
act=None,
name=''):
super(ResidualUnit, self).__init__()
self.if_shortcut = stride == 1 and in_channels == out_channels
self.if_se = use_se
self.expand_conv = ConvBNLayer(
in_channels=in_channels,
out_channels=mid_channels,
kernel_size=1,
stride=1,
padding=0,
if_act=True,
act=act,
name=name + "_expand")
self.bottleneck_conv = ConvBNLayer(
in_channels=mid_channels,
out_channels=mid_channels,
kernel_size=kernel_size,
stride=stride,
padding=int((kernel_size - 1) // 2),
groups=mid_channels,
if_act=True,
act=act,
name=name + "_depthwise")
if self.if_se:
self.mid_se = SEModule(mid_channels, name=name + "_se")
self.linear_conv = ConvBNLayer(
in_channels=mid_channels,
out_channels=out_channels,
kernel_size=1,
stride=1,
padding=0,
if_act=False,
act=None,
name=name + "_linear")
def forward(self, inputs):
x = self.expand_conv(inputs)
x = self.bottleneck_conv(x)
if self.if_se:
x = self.mid_se(x)
x = self.linear_conv(x)
if self.if_shortcut:
x = paddle.elementwise_add(inputs, x)
return x
class SEModule(nn.Layer):
def __init__(self, in_channels, reduction=4, name=""):
super(SEModule, self).__init__()
self.avg_pool = nn.Pool2D(
pool_type="avg", global_pooling=True, use_cudnn=False)
self.conv1 = nn.Conv2d(
in_channels=in_channels,
out_channels=in_channels // reduction,
kernel_size=1,
stride=1,
padding=0,
weight_attr=ParamAttr(name=name + "_1_weights"),
bias_attr=ParamAttr(name=name + "_1_offset"))
self.conv2 = nn.Conv2d(
in_channels=in_channels // reduction,
out_channels=in_channels,
kernel_size=1,
stride=1,
padding=0,
weight_attr=ParamAttr(name + "_2_weights"),
bias_attr=ParamAttr(name=name + "_2_offset"))
def forward(self, inputs):
outputs = self.avg_pool(inputs)
outputs = self.conv1(outputs)
outputs = F.relu(outputs)
outputs = self.conv2(outputs)
outputs = F.hard_sigmoid(outputs)
return inputs * outputs
ppocr/modeling/backbones/det_resnet_vd.py 0 → 100644
View file @ 4ffb5b62
# 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
from paddle import nn
from paddle.nn import functional as F
from paddle import ParamAttr
__all__ = ["ResNet"]
class ResNet(nn.Layer):
def __init__(self, in_channels=3, layers=50, **kwargs):
"""
the Resnet backbone network for detection module.
Args:
params(dict): the super parameters for network build
"""
super(ResNet, self).__init__()
supported_layers = {
18: {
'depth': [2, 2, 2, 2],
'block_class': BasicBlock
},
34: {
'depth': [3, 4, 6, 3],
'block_class': BasicBlock
},
50: {
'depth': [3, 4, 6, 3],
'block_class': BottleneckBlock
},
101: {
'depth': [3, 4, 23, 3],
'block_class': BottleneckBlock
},
152: {
'depth': [3, 8, 36, 3],
'block_class': BottleneckBlock
},
200: {
'depth': [3, 12, 48, 3],
'block_class': BottleneckBlock
}
}
assert layers in supported_layers, \
"supported layers are {} but input layer is {}".format(supported_layers.keys(), layers)
is_3x3 = True
depth = supported_layers[layers]['depth']
block_class = supported_layers[layers]['block_class']
num_filters = [64, 128, 256, 512]
conv = []
if is_3x3 == False:
conv.append(
ConvBNLayer(
in_channels=in_channels,
out_channels=64,
kernel_size=7,
stride=2,
act='relu'))
else:
conv.append(
ConvBNLayer(
in_channels=3,
out_channels=32,
kernel_size=3,
stride=2,
act='relu',
name='conv1_1'))
conv.append(
ConvBNLayer(
in_channels=32,
out_channels=32,
kernel_size=3,
stride=1,
act='relu',
name='conv1_2'))
conv.append(
ConvBNLayer(
in_channels=32,
out_channels=64,
kernel_size=3,
stride=1,
act='relu',
name='conv1_3'))
self.conv1 = nn.Sequential(*conv)
self.pool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
self.stages = []
self.out_channels = []
in_ch = 64
for block_index in range(len(depth)):
block_list = []
for i in range(depth[block_index]):
if layers >= 50:
if layers in [101, 152, 200] and block_index == 2:
if i == 0:
conv_name = "res" + str(block_index + 2) + "a"
else:
conv_name = "res" + str(block_index +
2) + "b" + str(i)
else:
conv_name = "res" + str(block_index + 2) + chr(97 + i)
else:
conv_name = "res" + str(block_index + 2) + chr(97 + i)
block_list.append(
block_class(
in_channels=in_ch,
out_channels=num_filters[block_index],
stride=2 if i == 0 and block_index != 0 else 1,
if_first=block_index == i == 0,
name=conv_name))
in_ch = block_list[-1].out_channels
self.out_channels.append(in_ch)
self.stages.append(nn.Sequential(*block_list))
for i, stage in enumerate(self.stages):
self.add_sublayer(sublayer=stage, name="stage{}".format(i))
def forward(self, x):
x = self.conv1(x)
x = self.pool(x)
out_list = []
for stage in self.stages:
x = stage(x)
out_list.append(x)
return out_list
class ConvBNLayer(nn.Layer):
def __init__(self,
in_channels,
out_channels,
kernel_size,
stride=1,
groups=1,
act=None,
name=None):
super(ConvBNLayer, self).__init__()
self.conv = nn.Conv2d(
in_channels=in_channels,
out_channels=out_channels,
kernel_size=kernel_size,
stride=stride,
padding=(kernel_size - 1) // 2,
groups=groups,
weight_attr=ParamAttr(name=name + "_weights"),
bias_attr=False)
if name == "conv1":
bn_name = "bn_" + name
else:
bn_name = "bn" + name[3:]
self.bn = nn.BatchNorm(
num_channels=out_channels,
act=act,
param_attr=ParamAttr(name=bn_name + "_scale"),
bias_attr=ParamAttr(name=bn_name + "_offset"),
moving_mean_name=bn_name + "_mean",
moving_variance_name=bn_name + "_variance")
def __call__(self, x):
x = self.conv(x)
x = self.bn(x)
return x
class ConvBNLayerNew(nn.Layer):
def __init__(self,
in_channels,
out_channels,
kernel_size,
stride=1,
groups=1,
act=None,
name=None):
super(ConvBNLayerNew, self).__init__()
self.pool = nn.AvgPool2d(
kernel_size=2, stride=2, padding=0, ceil_mode=True)
self.conv = nn.Conv2d(
in_channels=in_channels,
out_channels=out_channels,
kernel_size=kernel_size,
stride=1,
padding=(kernel_size - 1) // 2,
groups=groups,
weight_attr=ParamAttr(name=name + "_weights"),
bias_attr=False)
if name == "conv1":
bn_name = "bn_" + name
else:
bn_name = "bn" + name[3:]
self.bn = nn.BatchNorm(
num_channels=out_channels,
act=act,
param_attr=ParamAttr(name=bn_name + "_scale"),
bias_attr=ParamAttr(name=bn_name + "_offset"),
moving_mean_name=bn_name + "_mean",
moving_variance_name=bn_name + "_variance")
def __call__(self, x):
x = self.pool(x)
x = self.conv(x)
x = self.bn(x)
return x
class ShortCut(nn.Layer):
def __init__(self, in_channels, out_channels, stride, name, if_first=False):
super(ShortCut, self).__init__()
self.use_conv = True
if in_channels != out_channels or stride != 1:
if if_first:
self.conv = ConvBNLayer(
in_channels, out_channels, 1, stride, name=name)
else:
self.conv = ConvBNLayerNew(
in_channels, out_channels, 1, stride, name=name)
elif if_first:
self.conv = ConvBNLayer(
in_channels, out_channels, 1, stride, name=name)
else:
self.use_conv = False
def forward(self, x):
if self.use_conv:
x = self.conv(x)
return x
class BottleneckBlock(nn.Layer):
def __init__(self, in_channels, out_channels, stride, name, if_first):
super(BottleneckBlock, self).__init__()
self.conv0 = ConvBNLayer(
in_channels=in_channels,
out_channels=out_channels,
kernel_size=1,
act='relu',
name=name + "_branch2a")
self.conv1 = ConvBNLayer(
in_channels=out_channels,
out_channels=out_channels,
kernel_size=3,
stride=stride,
act='relu',
name=name + "_branch2b")
self.conv2 = ConvBNLayer(
in_channels=out_channels,
out_channels=out_channels * 4,
kernel_size=1,
act=None,
name=name + "_branch2c")
self.short = ShortCut(
in_channels=in_channels,
out_channels=out_channels * 4,
stride=stride,
if_first=if_first,
name=name + "_branch1")
self.out_channels = out_channels * 4
def forward(self, x):
y = self.conv0(x)
y = self.conv1(y)
y = self.conv2(y)
y = y + self.short(x)
y = F.relu(y)
return y
class BasicBlock(nn.Layer):
def __init__(self, in_channels, out_channels, stride, name, if_first):
super(BasicBlock, self).__init__()
self.conv0 = ConvBNLayer(
in_channels=in_channels,
out_channels=out_channels,
kernel_size=3,
act='relu',
stride=stride,
name=name + "_branch2a")
self.conv1 = ConvBNLayer(
in_channels=out_channels,
out_channels=out_channels,
kernel_size=3,
act=None,
name=name + "_branch2b")
self.short = ShortCut(
in_channels=in_channels,
out_channels=out_channels,
stride=stride,
if_first=if_first,
name=name + "_branch1")
self.out_channels = out_channels
def forward(self, x):
y = self.conv0(x)
y = self.conv1(y)
y = y + self.short(x)
return F.relu(y)
if __name__ == '__main__':
import paddle
paddle.disable_static()
x = paddle.zeros([1, 3, 640, 640])
x = paddle.to_variable(x)
print(x.shape)
net = ResNet(layers=18)
y = net(x)
for stage in y:
print(stage.shape)
# paddle.save(net.state_dict(),'1.pth')
ppocr/modeling/backbones/rec_mobilenet_v3.py 0 → 100644
View file @ 4ffb5b62
# 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 paddle import nn
from ppocr.modeling.backbones.det_mobilenet_v3 import ResidualUnit, ConvBNLayer, make_divisible
__all__ = ['MobileNetV3']
class MobileNetV3(nn.Layer):
def __init__(self,
in_channels=3,
model_name='small',
scale=0.5,
large_stride=None,
small_stride=None,
**kwargs):
super(MobileNetV3, self).__init__()
if small_stride is None:
small_stride = [2, 2, 2, 2]
if large_stride is None:
large_stride = [1, 2, 2, 2]
assert isinstance(large_stride, list), "large_stride type must " \
"be list but got {}".format(type(large_stride))
assert isinstance(small_stride, list), "small_stride type must " \
"be list but got {}".format(type(small_stride))
assert len(large_stride) == 4, "large_stride length must be " \
"4 but got {}".format(len(large_stride))
assert len(small_stride) == 4, "small_stride length must be " \
"4 but got {}".format(len(small_stride))
if model_name == "large":
cfg = [
# k, exp, c, se, nl, s,
[3, 16, 16, False, 'relu', large_stride[0]],
[3, 64, 24, False, 'relu', (large_stride[1], 1)],
[3, 72, 24, False, 'relu', 1],
[5, 72, 40, True, 'relu', (large_stride[2], 1)],
[5, 120, 40, True, 'relu', 1],
[5, 120, 40, True, 'relu', 1],
[3, 240, 80, False, 'hard_swish', 1],
[3, 200, 80, False, 'hard_swish', 1],
[3, 184, 80, False, 'hard_swish', 1],
[3, 184, 80, False, 'hard_swish', 1],
[3, 480, 112, True, 'hard_swish', 1],
[3, 672, 112, True, 'hard_swish', 1],
[5, 672, 160, True, 'hard_swish', (large_stride[3], 1)],
[5, 960, 160, True, 'hard_swish', 1],
[5, 960, 160, True, 'hard_swish', 1],
]
cls_ch_squeeze = 960
elif model_name == "small":
cfg = [
# k, exp, c, se, nl, s,
[3, 16, 16, True, 'relu', (small_stride[0], 1)],
[3, 72, 24, False, 'relu', (small_stride[1], 1)],
[3, 88, 24, False, 'relu', 1],
[5, 96, 40, True, 'hard_swish', (small_stride[2], 1)],
[5, 240, 40, True, 'hard_swish', 1],
[5, 240, 40, True, 'hard_swish', 1],
[5, 120, 48, True, 'hard_swish', 1],
[5, 144, 48, True, 'hard_swish', 1],
[5, 288, 96, True, 'hard_swish', (small_stride[3], 1)],
[5, 576, 96, True, 'hard_swish', 1],
[5, 576, 96, True, 'hard_swish', 1],
]
cls_ch_squeeze = 576
else:
raise NotImplementedError("mode[" + model_name +
"_model] is not implemented!")
supported_scale = [0.35, 0.5, 0.75, 1.0, 1.25]
assert scale in supported_scale, \
"supported scales are {} but input scale is {}".format(supported_scale, scale)
inplanes = 16
# conv1
self.conv1 = ConvBNLayer(
in_channels=in_channels,
out_channels=make_divisible(inplanes * scale),
kernel_size=3,
stride=2,
padding=1,
groups=1,
if_act=True,
act='hard_swish',
name='conv1')
i = 0
block_list = []
inplanes = make_divisible(inplanes * scale)
for (k, exp, c, se, nl, s) in cfg:
block_list.append(
ResidualUnit(
in_channels=inplanes,
mid_channels=make_divisible(scale * exp),
out_channels=make_divisible(scale * c),
kernel_size=k,
stride=s,
use_se=se,
act=nl,
name='conv' + str(i + 2)))
inplanes = make_divisible(scale * c)
i += 1
self.blocks = nn.Sequential(*block_list)
self.conv2 = ConvBNLayer(
in_channels=inplanes,
out_channels=make_divisible(scale * cls_ch_squeeze),
kernel_size=1,
stride=1,
padding=0,
groups=1,
if_act=True,
act='hard_swish',
name='conv_last')
self.pool = nn.MaxPool2d(kernel_size=2, stride=2, padding=0)
self.out_channels = make_divisible(scale * cls_ch_squeeze)
def forward(self, x):
x = self.conv1(x)
x = self.blocks(x)
x = self.conv2(x)
x = self.pool(x)
return x
if __name__ == '__main__':
import paddle
paddle.disable_static()
x = paddle.zeros((1, 3, 32, 320))
x = paddle.to_variable(x)
net = MobileNetV3(model_name='small', small_stride=[1, 2, 2, 2])
y = net(x)
print(y.shape)
ppocr/modeling/backbones/rec_resnet_vd.py 0 → 100644
View file @ 4ffb5b62
# 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
from paddle import nn, ParamAttr
from paddle.nn import functional as F
__all__ = ["ResNet"]
class ResNet(nn.Layer):
def __init__(self, in_channels=3, layers=34):
super(ResNet, self).__init__()
supported_layers = {
18: {
'depth': [2, 2, 2, 2],
'block_class': BasicBlock
},
34: {
'depth': [3, 4, 6, 3],
'block_class': BasicBlock
},
50: {
'depth': [3, 4, 6, 3],
'block_class': BottleneckBlock
},
101: {
'depth': [3, 4, 23, 3],
'block_class': BottleneckBlock
},
152: {
'depth': [3, 8, 36, 3],
'block_class': BottleneckBlock
},
200: {
'depth': [3, 12, 48, 3],
'block_class': BottleneckBlock
}
}
assert layers in supported_layers, \
"supported layers are {} but input layer is {}".format(supported_layers.keys(), layers)
is_3x3 = True
num_filters = [64, 128, 256, 512]
depth = supported_layers[layers]['depth']
block_class = supported_layers[layers]['block_class']
conv = []
if is_3x3 == False:
conv.append(
ConvBNLayer(
in_channels=in_channels,
out_channels=64,
kernel_size=7,
stride=1,
act='relu'))
else:
conv.append(
ConvBNLayer(
in_channels=in_channels,
out_channels=32,
kernel_size=3,
stride=1,
act='relu',
name='conv1_1'))
conv.append(
ConvBNLayer(
in_channels=32,
out_channels=32,
kernel_size=3,
stride=1,
act='relu',
name='conv1_2'))
conv.append(
ConvBNLayer(
in_channels=32,
out_channels=64,
kernel_size=3,
stride=1,
act='relu',
name='conv1_3'))
self.conv1 = nn.Sequential(*conv)
self.pool = nn.MaxPool2d(
kernel_size=3,
stride=2,
padding=1, )
block_list = []
in_ch = 64
for block_index in range(len(depth)):
for i in range(depth[block_index]):
if layers >= 50:
if layers in [101, 152, 200] and block_index == 2:
if i == 0:
conv_name = "res" + str(block_index + 2) + "a"
else:
conv_name = "res" + str(block_index +
2) + "b" + str(i)
else:
conv_name = "res" + str(block_index + 2) + chr(97 + i)
else:
conv_name = "res" + str(block_index + 2) + chr(97 + i)
if i == 0 and block_index != 0:
stride = (2, 1)
else:
stride = (1, 1)
block_list.append(
block_class(
in_channels=in_ch,
out_channels=num_filters[block_index],
stride=stride,
if_first=block_index == i == 0,
name=conv_name))
in_ch = block_list[-1].out_channels
self.block_list = nn.Sequential(*block_list)
self.add_sublayer(sublayer=self.block_list, name="block_list")
self.pool_out = nn.MaxPool2d(kernel_size=2, stride=2, padding=0)
self.out_channels = in_ch
def forward(self, x):
x = self.conv1(x)
x = self.pool(x)
x = self.block_list(x)
x = self.pool_out(x)
return x
class ConvBNLayer(nn.Layer):
def __init__(self,
in_channels,
out_channels,
kernel_size,
stride=1,
groups=1,
act=None,
name=None):
super(ConvBNLayer, self).__init__()
self.conv = nn.Conv2d(
in_channels=in_channels,
out_channels=out_channels,
kernel_size=kernel_size,
stride=stride,
padding=(kernel_size - 1) // 2,
groups=groups,
weight_attr=ParamAttr(name=name + "_weights"),
bias_attr=False)
if name == "conv1":
bn_name = "bn_" + name
else:
bn_name = "bn" + name[3:]
self.bn = nn.BatchNorm(
num_channels=out_channels,
act=act,
param_attr=ParamAttr(name=bn_name + "_scale"),
bias_attr=ParamAttr(name=bn_name + "_offset"),
moving_mean_name=bn_name + "_mean",
moving_variance_name=bn_name + "_variance")
def __call__(self, x):
x = self.conv(x)
x = self.bn(x)
return x
class ConvBNLayerNew(nn.Layer):
def __init__(self,
in_channels,
out_channels,
kernel_size,
stride=1,
groups=1,
act=None,
name=None):
super(ConvBNLayerNew, self).__init__()
self.pool = nn.AvgPool2d(
kernel_size=stride, stride=stride, padding=0, ceil_mode=True)
self.conv = nn.Conv2d(
in_channels=in_channels,
out_channels=out_channels,
kernel_size=kernel_size,
stride=1,
padding=(kernel_size - 1) // 2,
groups=groups,
weight_attr=ParamAttr(name=name + "_weights"),
bias_attr=False)
if name == "conv1":
bn_name = "bn_" + name
else:
bn_name = "bn" + name[3:]
self.bn = nn.BatchNorm(
num_channels=out_channels,
act=act,
param_attr=ParamAttr(name=bn_name + "_scale"),
bias_attr=ParamAttr(name=bn_name + "_offset"),
moving_mean_name=bn_name + "_mean",
moving_variance_name=bn_name + "_variance")
def __call__(self, x):
x = self.pool(x)
x = self.conv(x)
x = self.bn(x)
return x
class ShortCut(nn.Layer):
def __init__(self, in_channels, out_channels, stride, name, if_first=False):
super(ShortCut, self).__init__()
self.use_conv = True
if in_channels != out_channels or stride[0] != 1:
if if_first:
self.conv = ConvBNLayer(
in_channels, out_channels, 1, stride, name=name)
else:
self.conv = ConvBNLayerNew(
in_channels, out_channels, 1, stride, name=name)
elif if_first:
self.conv = ConvBNLayer(
in_channels, out_channels, 1, stride, name=name)
else:
self.use_conv = False
def forward(self, x):
if self.use_conv:
x = self.conv(x)
return x
class BottleneckBlock(nn.Layer):
def __init__(self, in_channels, out_channels, stride, name, if_first):
super(BottleneckBlock, self).__init__()
self.conv0 = ConvBNLayer(
in_channels=in_channels,
out_channels=out_channels,
kernel_size=1,
act='relu',
name=name + "_branch2a")
self.conv1 = ConvBNLayer(
in_channels=out_channels,
out_channels=out_channels,
kernel_size=3,
stride=stride,
act='relu',
name=name + "_branch2b")
self.conv2 = ConvBNLayer(
in_channels=out_channels,
out_channels=out_channels * 4,
kernel_size=1,
act=None,
name=name + "_branch2c")
self.short = ShortCut(
in_channels=in_channels,
out_channels=out_channels * 4,
stride=stride,
if_first=if_first,
name=name + "_branch1")
self.out_channels = out_channels * 4
def forward(self, x):
y = self.conv0(x)
y = self.conv1(y)
y = self.conv2(y)
y = y + self.short(x)
y = F.relu(y)
return y
class BasicBlock(nn.Layer):
def __init__(self, in_channels, out_channels, stride, name, if_first):
super(BasicBlock, self).__init__()
self.conv0 = ConvBNLayer(
in_channels=in_channels,
out_channels=out_channels,
kernel_size=3,
act='relu',
stride=stride,
name=name + "_branch2a")
self.conv1 = ConvBNLayer(
in_channels=out_channels,
out_channels=out_channels,
kernel_size=3,
act=None,
name=name + "_branch2b")
self.short = ShortCut(
in_channels=in_channels,
out_channels=out_channels,
stride=stride,
if_first=if_first,
name=name + "_branch1")
self.out_channels = out_channels
def forward(self, x):
y = self.conv0(x)
y = self.conv1(y)
y = y + self.short(x)
return F.relu(y)
ppocr/modeling/heads/__init__.py 0 → 100755
View file @ 4ffb5b62
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# 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.
__all__ = ['build_head']
def build_head(config):
# det head
from .det_db_head import DBHead
# rec head
from .rec_ctc_head import CTC
support_dict = ['DBHead', 'CTC']
module_name = config.pop('name')
assert module_name in support_dict, Exception('head only support {}'.format(
support_dict))
module_class = eval(module_name)(**config)
return module_class
ppocr/modeling/heads/det_db_head.py 0 → 100644
View file @ 4ffb5b62
# copyright (c) 2019 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
from paddle import nn
import paddle.nn.functional as F
from paddle import ParamAttr
def get_bias_attr(k, name):
stdv = 1.0 / math.sqrt(k * 1.0)
initializer = paddle.nn.initializer.Uniform(-stdv, stdv)
bias_attr = ParamAttr(initializer=initializer, name=name + "_b_attr")
return bias_attr
class Head(nn.Layer):
def __init__(self, in_channels, name_list):
super(Head, self).__init__()
self.conv1 = nn.Conv2d(
in_channels=in_channels,
out_channels=in_channels // 4,
kernel_size=3,
padding=1,
weight_attr=ParamAttr(name=name_list[0] + '.w_0'),
bias_attr=False)
self.conv_bn1 = nn.BatchNorm(
num_channels=in_channels // 4,
param_attr=ParamAttr(
name=name_list[1] + '.w_0',
initializer=paddle.nn.initializer.Constant(value=1.0)),
bias_attr=ParamAttr(
name=name_list[1] + '.b_0',
initializer=paddle.nn.initializer.Constant(value=1e-4)),
moving_mean_name=name_list[1] + '.w_1',
moving_variance_name=name_list[1] + '.w_2',
act='relu')
self.conv2 = nn.ConvTranspose2d(
in_channels=in_channels // 4,
out_channels=in_channels // 4,
kernel_size=2,
stride=2,
weight_attr=ParamAttr(
name=name_list[2] + '.w_0',
initializer=paddle.nn.initializer.MSRA(uniform=False)),
bias_attr=get_bias_attr(in_channels // 4, name_list[-1] + "conv2"))
self.conv_bn2 = nn.BatchNorm(
num_channels=in_channels // 4,
param_attr=ParamAttr(
name=name_list[3] + '.w_0',
initializer=paddle.nn.initializer.Constant(value=1.0)),
bias_attr=ParamAttr(
name=name_list[3] + '.b_0',
initializer=paddle.nn.initializer.Constant(value=1e-4)),
moving_mean_name=name_list[3] + '.w_1',
moving_variance_name=name_list[3] + '.w_2',
act="relu")
self.conv3 = nn.ConvTranspose2d(
in_channels=in_channels // 4,
out_channels=1,
kernel_size=2,
stride=2,
weight_attr=ParamAttr(
name=name_list[4] + '.w_0',
initializer=paddle.nn.initializer.MSRA(uniform=False)),
bias_attr=get_bias_attr(in_channels // 4, name_list[-1] + "conv3"),
)
def forward(self, x):
x = self.conv1(x)
x = self.conv_bn1(x)
x = self.conv2(x)
x = self.conv_bn2(x)
x = self.conv3(x)
x = F.sigmoid(x)
return x
class DBHead(nn.Layer):
"""
Differentiable Binarization (DB) for text detection:
see https://arxiv.org/abs/1911.08947
args:
params(dict): super parameters for build DB network
"""
def __init__(self, in_channels, k=50, **kwargs):
super(DBHead, self).__init__()
self.k = k
binarize_name_list = [
'conv2d_56', 'batch_norm_47', 'conv2d_transpose_0', 'batch_norm_48',
'conv2d_transpose_1', 'binarize'
]
thresh_name_list = [
'conv2d_57', 'batch_norm_49', 'conv2d_transpose_2', 'batch_norm_50',
'conv2d_transpose_3', 'thresh'
]
self.binarize = Head(in_channels, binarize_name_list)
self.thresh = Head(in_channels, thresh_name_list)
def step_function(self, x, y):
return paddle.reciprocal(1 + paddle.exp(-self.k * (x - y)))
def forward(self, x):
shrink_maps = self.binarize(x)
if not self.training:
return shrink_maps
threshold_maps = self.thresh(x)
binary_maps = self.step_function(shrink_maps, threshold_maps)
y = paddle.concat([shrink_maps, threshold_maps, binary_maps], axis=1)
return y
ppocr/modeling/heads/rec_ctc_head.py 0 → 100755
View file @ 4ffb5b62
# copyright (c) 2019 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
from paddle import ParamAttr, nn
def get_para_bias_attr(l2_decay, k, name):
regularizer = paddle.fluid.regularizer.L2Decay(l2_decay)
stdv = 1.0 / math.sqrt(k * 1.0)
initializer = nn.initializer.Uniform(-stdv, stdv)
weight_attr = ParamAttr(
regularizer=regularizer, initializer=initializer, name=name + "_w_attr")
bias_attr = ParamAttr(
regularizer=regularizer, initializer=initializer, name=name + "_b_attr")
return [weight_attr, bias_attr]
class CTC(nn.Layer):
def __init__(self, in_channels, out_channels, fc_decay=1e-5, **kwargs):
super(CTC, self).__init__()
weight_attr, bias_attr = get_para_bias_attr(
l2_decay=fc_decay, k=in_channels, name='ctc_fc')
self.fc = nn.Linear(
in_channels,
out_channels,
weight_attr=weight_attr,
bias_attr=bias_attr,
name='ctc_fc')
self.out_channels = out_channels
def forward(self, x, labels=None):
predicts = self.fc(x)
return predicts
ppocr/modeling/losses/__init__.py 0 → 100755
View file @ 4ffb5b62
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# 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.
import copy
def build_loss(config):
# det loss
from .det_db_loss import DBLoss
# rec loss
from .rec_ctc_loss import CTCLoss
support_dict = ['DBLoss', 'CTCLoss']
config = copy.deepcopy(config)
module_name = config.pop('name')
assert module_name in support_dict, Exception('loss only support {}'.format(
support_dict))
module_class = eval(module_name)(**config)
return module_class
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment