Skip to content

GitLab

Unverified Commit 26c16324 authored by Evezerest's avatar Evezerest Committed by GitHub
Browse files

Merge branch 'dygraph' into dygraph

parents d9549ce6 0b37c118
Hide whitespace changes
Inline Side-by-side
Showing with 249 additions and 109 deletions
ppocr/losses/rec_nrtr_loss.py
View file @ 26c16324
......@@ -22,7 +22,7 @@ class NRTRLoss(nn.Layer):
log_prb = F.log_softmax(pred, axis=1)
non_pad_mask = paddle.not_equal(
tgt, paddle.zeros(
tgt.shape, dtype='int64'))
tgt.shape, dtype=tgt.dtype))
loss = -(one_hot * log_prb).sum(axis=1)
loss = loss.masked_select(non_pad_mask).mean()
else:
......
ppocr/modeling/backbones/rec_mv1_enhance.py
View file @ 26c16324
......@@ -12,6 +12,8 @@
# See the License for the specific language governing permissions and
# limitations under the License.
# This code is refer from: https://github.com/PaddlePaddle/PaddleClas/blob/develop/ppcls/arch/backbone/legendary_models/pp_lcnet.py
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
......
ppocr/modeling/backbones/rec_resnet_31.py
View file @ 26c16324
# copyright (c) 2021 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.
"""
This code is refer from:
https://github.com/open-mmlab/mmocr/blob/main/mmocr/models/textrecog/layers/conv_layer.py
https://github.com/open-mmlab/mmocr/blob/main/mmocr/models/textrecog/backbones/resnet31_ocr.py
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
......@@ -18,12 +37,12 @@ def conv3x3(in_channel, out_channel, stride=1):
kernel_size=3,
stride=stride,
padding=1,
bias_attr=False
)
bias_attr=False)
class BasicBlock(nn.Layer):
expansion = 1
def __init__(self, in_channels, channels, stride=1, downsample=False):
super().__init__()
self.conv1 = conv3x3(in_channels, channels, stride)
......@@ -34,9 +53,13 @@ class BasicBlock(nn.Layer):
self.downsample = downsample
if downsample:
self.downsample = nn.Sequential(
nn.Conv2D(in_channels, channels * self.expansion, 1, stride, bias_attr=False),
nn.BatchNorm2D(channels * self.expansion),
)
nn.Conv2D(
in_channels,
channels * self.expansion,
1,
stride,
bias_attr=False),
nn.BatchNorm2D(channels * self.expansion), )
else:
self.downsample = nn.Sequential()
self.stride = stride
......@@ -57,7 +80,7 @@ class BasicBlock(nn.Layer):
out += residual
out = self.relu(out)
return out
return out
class ResNet31(nn.Layer):
......@@ -69,12 +92,13 @@ class ResNet31(nn.Layer):
out_indices (None | Sequence[int]): Indices of output stages.
last_stage_pool (bool): If True, add `MaxPool2d` layer to last stage.
'''
def __init__(self,
in_channels=3,
layers=[1, 2, 5, 3],
channels=[64, 128, 256, 256, 512, 512, 512],
out_indices=None,
last_stage_pool=False):
def __init__(self,
in_channels=3,
layers=[1, 2, 5, 3],
channels=[64, 128, 256, 256, 512, 512, 512],
out_indices=None,
last_stage_pool=False):
super(ResNet31, self).__init__()
assert isinstance(in_channels, int)
assert isinstance(last_stage_pool, bool)
......@@ -83,46 +107,56 @@ class ResNet31(nn.Layer):
self.last_stage_pool = last_stage_pool
# conv 1 (Conv Conv)
self.conv1_1 = nn.Conv2D(in_channels, channels[0], kernel_size=3, stride=1, padding=1)
self.conv1_1 = nn.Conv2D(
in_channels, channels[0], kernel_size=3, stride=1, padding=1)
self.bn1_1 = nn.BatchNorm2D(channels[0])
self.relu1_1 = nn.ReLU()
self.conv1_2 = nn.Conv2D(channels[0], channels[1], kernel_size=3, stride=1, padding=1)
self.conv1_2 = nn.Conv2D(
channels[0], channels[1], kernel_size=3, stride=1, padding=1)
self.bn1_2 = nn.BatchNorm2D(channels[1])
self.relu1_2 = nn.ReLU()
# conv 2 (Max-pooling, Residual block, Conv)
self.pool2 = nn.MaxPool2D(kernel_size=2, stride=2, padding=0, ceil_mode=True)
self.pool2 = nn.MaxPool2D(
kernel_size=2, stride=2, padding=0, ceil_mode=True)
self.block2 = self._make_layer(channels[1], channels[2], layers[0])
self.conv2 = nn.Conv2D(channels[2], channels[2], kernel_size=3, stride=1, padding=1)
self.conv2 = nn.Conv2D(
channels[2], channels[2], kernel_size=3, stride=1, padding=1)
self.bn2 = nn.BatchNorm2D(channels[2])
self.relu2 = nn.ReLU()
# conv 3 (Max-pooling, Residual block, Conv)
self.pool3 = nn.MaxPool2D(kernel_size=2, stride=2, padding=0, ceil_mode=True)
self.pool3 = nn.MaxPool2D(
kernel_size=2, stride=2, padding=0, ceil_mode=True)
self.block3 = self._make_layer(channels[2], channels[3], layers[1])
self.conv3 = nn.Conv2D(channels[3], channels[3], kernel_size=3, stride=1, padding=1)
self.conv3 = nn.Conv2D(
channels[3], channels[3], kernel_size=3, stride=1, padding=1)
self.bn3 = nn.BatchNorm2D(channels[3])
self.relu3 = nn.ReLU()
# conv 4 (Max-pooling, Residual block, Conv)
self.pool4 = nn.MaxPool2D(kernel_size=(2, 1), stride=(2, 1), padding=0, ceil_mode=True)
self.pool4 = nn.MaxPool2D(
kernel_size=(2, 1), stride=(2, 1), padding=0, ceil_mode=True)
self.block4 = self._make_layer(channels[3], channels[4], layers[2])
self.conv4 = nn.Conv2D(channels[4], channels[4], kernel_size=3, stride=1, padding=1)
self.conv4 = nn.Conv2D(
channels[4], channels[4], kernel_size=3, stride=1, padding=1)
self.bn4 = nn.BatchNorm2D(channels[4])
self.relu4 = nn.ReLU()
# conv 5 ((Max-pooling), Residual block, Conv)
self.pool5 = None
if self.last_stage_pool:
self.pool5 = nn.MaxPool2D(kernel_size=2, stride=2, padding=0, ceil_mode=True)
self.pool5 = nn.MaxPool2D(
kernel_size=2, stride=2, padding=0, ceil_mode=True)
self.block5 = self._make_layer(channels[4], channels[5], layers[3])
self.conv5 = nn.Conv2D(channels[5], channels[5], kernel_size=3, stride=1, padding=1)
self.conv5 = nn.Conv2D(
channels[5], channels[5], kernel_size=3, stride=1, padding=1)
self.bn5 = nn.BatchNorm2D(channels[5])
self.relu5 = nn.ReLU()
self.out_channels = channels[-1]
def _make_layer(self, input_channels, output_channels, blocks):
layers = []
for _ in range(blocks):
......@@ -130,19 +164,19 @@ class ResNet31(nn.Layer):
if input_channels != output_channels:
downsample = nn.Sequential(
nn.Conv2D(
input_channels,
output_channels,
kernel_size=1,
stride=1,
input_channels,
output_channels,
kernel_size=1,
stride=1,
bias_attr=False),
nn.BatchNorm2D(output_channels),
)
layers.append(BasicBlock(input_channels, output_channels, downsample=downsample))
nn.BatchNorm2D(output_channels), )
layers.append(
BasicBlock(
input_channels, output_channels, downsample=downsample))
input_channels = output_channels
return nn.Sequential(*layers)
def forward(self, x):
x = self.conv1_1(x)
x = self.bn1_1(x)
......@@ -166,11 +200,11 @@ class ResNet31(nn.Layer):
x = block_layer(x)
x = conv_layer(x)
x = bn_layer(x)
x= relu_layer(x)
x = relu_layer(x)
outs.append(x)
if self.out_indices is not None:
return tuple([outs[i] for i in self.out_indices])
return x
ppocr/modeling/backbones/rec_resnet_aster.py
View file @ 26c16324
......@@ -11,7 +11,10 @@
# 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.
"""
This code is refer from:
https://github.com/ayumiymk/aster.pytorch/blob/master/lib/models/resnet_aster.py
"""
import paddle
import paddle.nn as nn
......
ppocr/modeling/heads/det_pse_head.py
View file @ 26c16324
# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
# copyright (c) 2021 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.
......@@ -11,22 +11,24 @@
# 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.
"""
This code is refer from:
https://github.com/whai362/PSENet/blob/python3/models/head/psenet_head.py
"""
from paddle import nn
class PSEHead(nn.Layer):
def __init__(self,
in_channels,
hidden_dim=256,
out_channels=7,
**kwargs):
def __init__(self, in_channels, hidden_dim=256, out_channels=7, **kwargs):
super(PSEHead, self).__init__()
self.conv1 = nn.Conv2D(in_channels, hidden_dim, kernel_size=3, stride=1, padding=1)
self.conv1 = nn.Conv2D(
in_channels, hidden_dim, kernel_size=3, stride=1, padding=1)
self.bn1 = nn.BatchNorm2D(hidden_dim)
self.relu1 = nn.ReLU()
self.conv2 = nn.Conv2D(hidden_dim, out_channels, kernel_size=1, stride=1, padding=0)
self.conv2 = nn.Conv2D(
hidden_dim, out_channels, kernel_size=1, stride=1, padding=0)
def forward(self, x, **kwargs):
out = self.conv1(x)
......
ppocr/modeling/heads/rec_aster_head.py
View file @ 26c16324
......@@ -11,6 +11,10 @@
# 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.
"""
This code is refer from:
https://github.com/ayumiymk/aster.pytorch/blob/master/lib/models/attention_recognition_head.py
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
......
ppocr/modeling/heads/rec_att_head.py
View file @ 26c16324
......@@ -75,7 +75,7 @@ class AttentionHead(nn.Layer):
probs_step, axis=1)], axis=1)
next_input = probs_step.argmax(axis=1)
targets = next_input
probs = paddle.nn.functional.softmax(probs, axis=2)
return probs
......
ppocr/modeling/heads/rec_sar_head.py
View file @ 26c16324
# copyright (c) 2021 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.
"""
This code is refer from:
https://github.com/open-mmlab/mmocr/blob/main/mmocr/models/textrecog/encoders/sar_encoder.py
https://github.com/open-mmlab/mmocr/blob/main/mmocr/models/textrecog/decoders/sar_decoder.py
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
......@@ -275,7 +294,6 @@ class ParallelSARDecoder(BaseDecoder):
if img_metas is not None and self.mask:
valid_ratios = img_metas[-1]
label = label.cuda()
lab_embedding = self.embedding(label)
# bsz * seq_len * emb_dim
out_enc = out_enc.unsqueeze(1)
......
ppocr/modeling/necks/fpn.py
View file @ 26c16324
......@@ -11,64 +11,102 @@
# 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.
"""
This code is refer from:
https://github.com/whai362/PSENet/blob/python3/models/neck/fpn.py
"""
import paddle.nn as nn
import paddle
import math
import paddle.nn.functional as F
class Conv_BN_ReLU(nn.Layer):
def __init__(self, in_planes, out_planes, kernel_size=1, stride=1, padding=0):
def __init__(self,
in_planes,
out_planes,
kernel_size=1,
stride=1,
padding=0):
super(Conv_BN_ReLU, self).__init__()
self.conv = nn.Conv2D(in_planes, out_planes, kernel_size=kernel_size, stride=stride, padding=padding,
bias_attr=False)
self.conv = nn.Conv2D(
in_planes,
out_planes,
kernel_size=kernel_size,
stride=stride,
padding=padding,
bias_attr=False)
self.bn = nn.BatchNorm2D(out_planes, momentum=0.1)
self.relu = nn.ReLU()
for m in self.sublayers():
if isinstance(m, nn.Conv2D):
n = m._kernel_size[0] * m._kernel_size[1] * m._out_channels
m.weight = paddle.create_parameter(shape=m.weight.shape, dtype='float32', default_initializer=paddle.nn.initializer.Normal(0, math.sqrt(2. / n)))
m.weight = paddle.create_parameter(
shape=m.weight.shape,
dtype='float32',
default_initializer=paddle.nn.initializer.Normal(
0, math.sqrt(2. / n)))
elif isinstance(m, nn.BatchNorm2D):
m.weight = paddle.create_parameter(shape=m.weight.shape, dtype='float32', default_initializer=paddle.nn.initializer.Constant(1.0))
m.bias = paddle.create_parameter(shape=m.bias.shape, dtype='float32', default_initializer=paddle.nn.initializer.Constant(0.0))
m.weight = paddle.create_parameter(
shape=m.weight.shape,
dtype='float32',
default_initializer=paddle.nn.initializer.Constant(1.0))
m.bias = paddle.create_parameter(
shape=m.bias.shape,
dtype='float32',
default_initializer=paddle.nn.initializer.Constant(0.0))
def forward(self, x):
return self.relu(self.bn(self.conv(x)))
class FPN(nn.Layer):
def __init__(self, in_channels, out_channels):
super(FPN, self).__init__()
# Top layer
self.toplayer_ = Conv_BN_ReLU(in_channels[3], out_channels, kernel_size=1, stride=1, padding=0)
self.toplayer_ = Conv_BN_ReLU(
in_channels[3], out_channels, kernel_size=1, stride=1, padding=0)
# Lateral layers
self.latlayer1_ = Conv_BN_ReLU(in_channels[2], out_channels, kernel_size=1, stride=1, padding=0)
self.latlayer1_ = Conv_BN_ReLU(
in_channels[2], out_channels, kernel_size=1, stride=1, padding=0)
self.latlayer2_ = Conv_BN_ReLU(in_channels[1], out_channels, kernel_size=1, stride=1, padding=0)
self.latlayer2_ = Conv_BN_ReLU(
in_channels[1], out_channels, kernel_size=1, stride=1, padding=0)
self.latlayer3_ = Conv_BN_ReLU(in_channels[0], out_channels, kernel_size=1, stride=1, padding=0)
self.latlayer3_ = Conv_BN_ReLU(
in_channels[0], out_channels, kernel_size=1, stride=1, padding=0)
# Smooth layers
self.smooth1_ = Conv_BN_ReLU(out_channels, out_channels, kernel_size=3, stride=1, padding=1)
self.smooth2_ = Conv_BN_ReLU(out_channels, out_channels, kernel_size=3, stride=1, padding=1)
self.smooth1_ = Conv_BN_ReLU(
out_channels, out_channels, kernel_size=3, stride=1, padding=1)
self.smooth3_ = Conv_BN_ReLU(out_channels, out_channels, kernel_size=3, stride=1, padding=1)
self.smooth2_ = Conv_BN_ReLU(
out_channels, out_channels, kernel_size=3, stride=1, padding=1)
self.smooth3_ = Conv_BN_ReLU(
out_channels, out_channels, kernel_size=3, stride=1, padding=1)
self.out_channels = out_channels * 4
for m in self.sublayers():
if isinstance(m, nn.Conv2D):
n = m._kernel_size[0] * m._kernel_size[1] * m._out_channels
m.weight = paddle.create_parameter(shape=m.weight.shape, dtype='float32',
default_initializer=paddle.nn.initializer.Normal(0,
math.sqrt(2. / n)))
m.weight = paddle.create_parameter(
shape=m.weight.shape,
dtype='float32',
default_initializer=paddle.nn.initializer.Normal(
0, math.sqrt(2. / n)))
elif isinstance(m, nn.BatchNorm2D):
m.weight = paddle.create_parameter(shape=m.weight.shape, dtype='float32',
default_initializer=paddle.nn.initializer.Constant(1.0))
m.bias = paddle.create_parameter(shape=m.bias.shape, dtype='float32',
default_initializer=paddle.nn.initializer.Constant(0.0))
m.weight = paddle.create_parameter(
shape=m.weight.shape,
dtype='float32',
default_initializer=paddle.nn.initializer.Constant(1.0))
m.bias = paddle.create_parameter(
shape=m.bias.shape,
dtype='float32',
default_initializer=paddle.nn.initializer.Constant(0.0))
def _upsample(self, x, scale=1):
return F.upsample(x, scale_factor=scale, mode='bilinear')
......@@ -81,15 +119,15 @@ class FPN(nn.Layer):
p5 = self.toplayer_(f5)
f4 = self.latlayer1_(f4)
p4 = self._upsample_add(p5, f4,2)
p4 = self._upsample_add(p5, f4, 2)
p4 = self.smooth1_(p4)
f3 = self.latlayer2_(f3)
p3 = self._upsample_add(p4, f3,2)
p3 = self._upsample_add(p4, f3, 2)
p3 = self.smooth2_(p3)
f2 = self.latlayer3_(f2)
p2 = self._upsample_add(p3, f2,2)
p2 = self._upsample_add(p3, f2, 2)
p2 = self.smooth3_(p2)
p3 = self._upsample(p3, 2)
......@@ -97,4 +135,4 @@ class FPN(nn.Layer):
p5 = self._upsample(p5, 8)
fuse = paddle.concat([p2, p3, p4, p5], axis=1)
return fuse
\ No newline at end of file
return fuse
ppocr/modeling/transforms/stn.py
View file @ 26c16324
......@@ -11,7 +11,10 @@
# 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.
"""
This code is refer from:
https://github.com/ayumiymk/aster.pytorch/blob/master/lib/models/stn_head.py
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
......
ppocr/modeling/transforms/tps.py
View file @ 26c16324
......@@ -11,6 +11,10 @@
# 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.
"""
This code is refer from:
https://github.com/clovaai/deep-text-recognition-benchmark/blob/master/modules/transformation.py
"""
from __future__ import absolute_import
from __future__ import division
......
ppocr/modeling/transforms/tps_spatial_transformer.py
View file @ 26c16324
......@@ -11,6 +11,10 @@
# 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.
"""
This code is refer from:
https://github.com/ayumiymk/aster.pytorch/blob/master/lib/models/tps_spatial_transformer.py
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
......
ppocr/postprocess/__init__.py
View file @ 26c16324
......@@ -18,7 +18,6 @@ from __future__ import print_function
from __future__ import unicode_literals
import copy
import platform
__all__ = ['build_post_process']
......@@ -26,24 +25,24 @@ from .db_postprocess import DBPostProcess, DistillationDBPostProcess
from .east_postprocess import EASTPostProcess
from .sast_postprocess import SASTPostProcess
from .rec_postprocess import CTCLabelDecode, AttnLabelDecode, SRNLabelDecode, DistillationCTCLabelDecode, \
TableLabelDecode, NRTRLabelDecode, SARLabelDecode , SEEDLabelDecode
TableLabelDecode, NRTRLabelDecode, SARLabelDecode, SEEDLabelDecode
from .cls_postprocess import ClsPostProcess
from .pg_postprocess import PGPostProcess
if platform.system() != "Windows":
# pse is not support in Windows
from .pse_postprocess import PSEPostProcess
def build_post_process(config, global_config=None):
support_dict = [
'DBPostProcess', 'PSEPostProcess', 'EASTPostProcess', 'SASTPostProcess',
'CTCLabelDecode', 'AttnLabelDecode', 'ClsPostProcess', 'SRNLabelDecode',
'PGPostProcess', 'DistillationCTCLabelDecode', 'TableLabelDecode',
'DBPostProcess', 'EASTPostProcess', 'SASTPostProcess', 'CTCLabelDecode',
'AttnLabelDecode', 'ClsPostProcess', 'SRNLabelDecode', 'PGPostProcess',
'DistillationCTCLabelDecode', 'TableLabelDecode',
'DistillationDBPostProcess', 'NRTRLabelDecode', 'SARLabelDecode',
'SEEDLabelDecode'
]
if config['name'] == 'PSEPostProcess':
from .pse_postprocess import PSEPostProcess
support_dict.append('PSEPostProcess')
config = copy.deepcopy(config)
module_name = config.pop('name')
if global_config is not None:
......
ppocr/postprocess/db_postprocess.py
View file @ 26c16324
......@@ -11,7 +11,10 @@
# 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.
"""
This code is refered from:
https://github.com/WenmuZhou/DBNet.pytorch/blob/master/post_processing/seg_detector_representer.py
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
......@@ -190,7 +193,8 @@ class DBPostProcess(object):
class DistillationDBPostProcess(object):
def __init__(self, model_name=["student"],
def __init__(self,
model_name=["student"],
key=None,
thresh=0.3,
box_thresh=0.6,
......@@ -201,12 +205,13 @@ class DistillationDBPostProcess(object):
**kwargs):
self.model_name = model_name
self.key = key
self.post_process = DBPostProcess(thresh=thresh,
box_thresh=box_thresh,
max_candidates=max_candidates,
unclip_ratio=unclip_ratio,
use_dilation=use_dilation,
score_mode=score_mode)
self.post_process = DBPostProcess(
thresh=thresh,
box_thresh=box_thresh,
max_candidates=max_candidates,
unclip_ratio=unclip_ratio,
use_dilation=use_dilation,
score_mode=score_mode)
def __call__(self, predicts, shape_list):
results = {}
......
ppocr/postprocess/locality_aware_nms.py
View file @ 26c16324
"""
Locality aware nms.
This code is refered from: https://github.com/songdejia/EAST/blob/master/locality_aware_nms.py
"""
import numpy as np
......
ppocr/postprocess/pse_postprocess/pse/README.md
View file @ 26c16324
## 编译
code from https://github.com/whai362/pan_pp.pytorch
This code is refer from:
https://github.com/whai362/PSENet/blob/python3/models/post_processing/pse
```python
python3 setup.py build_ext --inplace
```
ppocr/postprocess/pse_postprocess/pse/__init__.py
View file @ 26c16324
......@@ -17,7 +17,13 @@ import subprocess
python_path = sys.executable
if subprocess.call('cd ppocr/postprocess/pse_postprocess/pse;{} setup.py build_ext --inplace;cd -'.format(python_path), shell=True) != 0:
raise RuntimeError('Cannot compile pse: {}'.format(os.path.dirname(os.path.realpath(__file__))))
ori_path = os.getcwd()
os.chdir('ppocr/postprocess/pse_postprocess/pse')
if subprocess.call(
'{} setup.py build_ext --inplace'.format(python_path), shell=True) != 0:
raise RuntimeError(
'Cannot compile pse: {}, if your system is windows, you need to install all the default components of `desktop development using C++` in visual studio 2019+'.
format(os.path.dirname(os.path.realpath(__file__))))
os.chdir(ori_path)
from .pse import pse
\ No newline at end of file
from .pse import pse
ppocr/postprocess/pse_postprocess/pse_postprocess.py
View file @ 26c16324
# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
# copyright (c) 2021 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
# 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.
"""
This code is refer from:
https://github.com/whai362/PSENet/blob/python3/models/head/psenet_head.py
"""
from __future__ import absolute_import
from __future__ import division
......@@ -47,7 +51,8 @@ class PSEPostProcess(object):
pred = outs_dict['maps']
if not isinstance(pred, paddle.Tensor):
pred = paddle.to_tensor(pred)
pred = F.interpolate(pred, scale_factor=4 // self.scale, mode='bilinear')
pred = F.interpolate(
pred, scale_factor=4 // self.scale, mode='bilinear')
score = F.sigmoid(pred[:, 0, :, :])
......@@ -60,7 +65,9 @@ class PSEPostProcess(object):
boxes_batch = []
for batch_index in range(pred.shape[0]):
boxes, scores = self.boxes_from_bitmap(score[batch_index], kernels[batch_index], shape_list[batch_index])
boxes, scores = self.boxes_from_bitmap(score[batch_index],
kernels[batch_index],
shape_list[batch_index])
boxes_batch.append({'points': boxes, 'scores': scores})
return boxes_batch
......@@ -98,15 +105,14 @@ class PSEPostProcess(object):
mask = np.zeros((box_height, box_width), np.uint8)
mask[points[:, 1], points[:, 0]] = 255
contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_SIMPLE)
bbox = np.squeeze(contours[0], 1)
else:
raise NotImplementedError
bbox[:, 0] = np.clip(
np.round(bbox[:, 0] / ratio_w), 0, src_w)
bbox[:, 1] = np.clip(
np.round(bbox[:, 1] / ratio_h), 0, src_h)
bbox[:, 0] = np.clip(np.round(bbox[:, 0] / ratio_w), 0, src_w)
bbox[:, 1] = np.clip(np.round(bbox[:, 1] / ratio_h), 0, src_h)
boxes.append(bbox)
scores.append(score_i)
return boxes, scores
ppocr/utils/iou.py
View file @ 26c16324
# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
# copyright (c) 2021 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.
......@@ -11,18 +11,23 @@
# 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.
"""
This code is refer from:
https://github.com/whai362/PSENet/blob/python3/models/loss/iou.py
"""
import paddle
EPS = 1e-6
def iou_single(a, b, mask, n_class):
valid = mask == 1
a = a.masked_select(valid)
b = b.masked_select(valid)
miou = []
for i in range(n_class):
if a.shape == [0] and a.shape==b.shape:
if a.shape == [0] and a.shape == b.shape:
inter = paddle.to_tensor(0.0)
union = paddle.to_tensor(0.0)
else:
......@@ -32,6 +37,7 @@ def iou_single(a, b, mask, n_class):
miou = sum(miou) / len(miou)
return miou
def iou(a, b, mask, n_class=2, reduce=True):
batch_size = a.shape[0]
......@@ -39,10 +45,10 @@ def iou(a, b, mask, n_class=2, reduce=True):
b = b.reshape([batch_size, -1])
mask = mask.reshape([batch_size, -1])
iou = paddle.zeros((batch_size,), dtype='float32')
iou = paddle.zeros((batch_size, ), dtype='float32')
for i in range(batch_size):
iou[i] = iou_single(a[i], b[i], mask[i], n_class)
if reduce:
iou = paddle.mean(iou)
return iou
\ No newline at end of file
return iou
ppocr/utils/logging.py
View file @ 26c16324
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
# 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.
......@@ -11,6 +11,10 @@
# 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.
"""
This code is refer from:
https://github.com/WenmuZhou/PytorchOCR/blob/master/torchocr/utils/logging.py
"""
import os
import sys
......
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