diff --git a/LICENSE b/LICENSE
index 261eeb9e9f..5fe86943b3 100644
--- a/LICENSE
+++ b/LICENSE
@@ -1,3 +1,5 @@
+Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved
+
Apache License
Version 2.0, January 2004
http://www.apache.org/licenses/
@@ -186,7 +188,7 @@
same "printed page" as the copyright notice for easier
identification within third-party archives.
- Copyright [yyyy] [name of copyright owner]
+ Copyright (c) 2016 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.
diff --git a/PPOCRLabel/PPOCRLabel.py b/PPOCRLabel/PPOCRLabel.py
index 0a3ae1cb3b..6c8154d1c9 100644
--- a/PPOCRLabel/PPOCRLabel.py
+++ b/PPOCRLabel/PPOCRLabel.py
@@ -1617,8 +1617,9 @@ def showBoundingBoxFromPPlabel(self, filePath):
key_cls = 'None' if not self.kie_mode else box.get('key_cls', 'None')
shapes.append((box['transcription'], box['points'], None, key_cls, box.get('difficult', False)))
- self.loadLabels(shapes)
- self.canvas.verified = False
+ if shapes != []:
+ self.loadLabels(shapes)
+ self.canvas.verified = False
def validFilestate(self, filePath):
if filePath not in self.fileStatedict.keys():
@@ -2203,7 +2204,7 @@ def reRecognition(self):
msg = 'Can not recognise the detection box in ' + self.filePath + '. Please change manually'
QMessageBox.information(self, "Information", msg)
return
- result = self.ocr.ocr(img_crop, cls=True, det=False)
+ result = self.ocr.ocr(img_crop, cls=True, det=False)[0]
if result[0][0] != '':
if shape.line_color == DEFAULT_LOCK_COLOR:
shape.label = result[0][0]
@@ -2264,7 +2265,7 @@ def singleRerecognition(self):
msg = 'Can not recognise the detection box in ' + self.filePath + '. Please change manually'
QMessageBox.information(self, "Information", msg)
return
- result = self.ocr.ocr(img_crop, cls=True, det=False)
+ result = self.ocr.ocr(img_crop, cls=True, det=False)[0]
if result[0][0] != '':
result.insert(0, box)
print('result in reRec is ', result)
@@ -2415,12 +2416,12 @@ def cellreRecognition(self):
# merge the text result in the cell
texts = ''
probs = 0. # the probability of the cell is avgerage prob of every text box in the cell
- bboxes = self.ocr.ocr(img_crop, det=True, rec=False, cls=False)
+ bboxes = self.ocr.ocr(img_crop, det=True, rec=False, cls=False)[0]
if len(bboxes) > 0:
bboxes.reverse() # top row text at first
for _bbox in bboxes:
patch = get_rotate_crop_image(img_crop, np.array(_bbox, np.float32))
- rec_res = self.ocr.ocr(patch, det=False, rec=True, cls=False)
+ rec_res = self.ocr.ocr(patch, det=False, rec=True, cls=False)[0]
text = rec_res[0][0]
if text != '':
texts += text + ('' if text[0].isalpha() else ' ') # add space between english word
diff --git a/PPOCRLabel/README.md b/PPOCRLabel/README.md
index 9c483e1fef..ec933a8397 100644
--- a/PPOCRLabel/README.md
+++ b/PPOCRLabel/README.md
@@ -103,11 +103,11 @@ python PPOCRLabel.py --kie True # [KIE mode] for [detection + recognition + keyw
```
#### 1.2.3 Build and Install the Whl Package Locally
-Compile and install a new whl package, where 1.0.2 is the version number, you can specify the new version in 'setup.py'.
+Compile and install a new whl package, where 0.0.0 is the version number, you can specify the new version in 'setup.py'.
```bash
cd ./PPOCRLabel
python3 setup.py bdist_wheel
-pip3 install dist/PPOCRLabel-2.1.2-py2.py3-none-any.whl
+pip3 install dist/PPOCRLabel-0.0.0-py2.py3-none-any.whl
```
diff --git a/PPOCRLabel/README_ch.md b/PPOCRLabel/README_ch.md
index afe1a08ff9..5359afc4f1 100644
--- a/PPOCRLabel/README_ch.md
+++ b/PPOCRLabel/README_ch.md
@@ -101,12 +101,12 @@ python PPOCRLabel.py --lang ch
#### 1.2.3 本地构建whl包并安装
-编译与安装新的whl包，其中1.0.2为版本号，可在 `setup.py` 中指定新版本。
+编译与安装新的whl包，其中0.0.0为版本号，可在 `setup.py` 中指定新版本。
```bash
cd ./PPOCRLabel
python3 setup.py bdist_wheel
-pip3 install dist/PPOCRLabel-2.1.2-py2.py3-none-any.whl -i https://mirror.baidu.com/pypi/simple
+pip3 install dist/PPOCRLabel-0.0.0-py2.py3-none-any.whl -i https://mirror.baidu.com/pypi/simple
```
diff --git a/PPOCRLabel/libs/autoDialog.py b/PPOCRLabel/libs/autoDialog.py
index 189a590de8..55636eec0f 100644
--- a/PPOCRLabel/libs/autoDialog.py
+++ b/PPOCRLabel/libs/autoDialog.py
@@ -40,7 +40,7 @@ def run(self):
if self.model == 'paddle':
h, w, _ = cv2.imdecode(np.fromfile(Imgpath, dtype=np.uint8), 1).shape
if h > 32 and w > 32:
- self.result_dic = self.ocr.ocr(Imgpath, cls=True, det=True)
+ self.result_dic = self.ocr.ocr(Imgpath, cls=True, det=True)[0]
else:
print('The size of', Imgpath, 'is too small to be recognised')
self.result_dic = None
diff --git a/PPOCRLabel/requirements.txt b/PPOCRLabel/requirements.txt
index fd42a2de25..a10b3453a9 100644
--- a/PPOCRLabel/requirements.txt
+++ b/PPOCRLabel/requirements.txt
@@ -1,3 +1,3 @@
pyqt5
-paddleocr==2.6.0.0
+paddleocr
xlrd==1.2.0
diff --git a/PPOCRLabel/setup.py b/PPOCRLabel/setup.py
index a112df544e..9770b632bd 100644
--- a/PPOCRLabel/setup.py
+++ b/PPOCRLabel/setup.py
@@ -33,10 +33,10 @@ def readme():
package_dir={'PPOCRLabel': ''},
include_package_data=True,
entry_points={"console_scripts": ["PPOCRLabel= PPOCRLabel.PPOCRLabel:main"]},
- version='2.1.2',
+ version='2.1.3',
install_requires=requirements,
license='Apache License 2.0',
- description='PPOCRLabel is a semi-automatic graphic annotation tool suitable for OCR field, with built-in PPOCR model to automatically detect and re-recognize data. It is written in python3 and pyqt5, supporting rectangular box annotation and four-point annotation modes. Annotations can be directly used for the training of PPOCR detection and recognition models',
+ description='PPOCRLabelv2 is a semi-automatic graphic annotation tool suitable for OCR field, with built-in PP-OCR model to automatically detect and re-recognize data. It is written in Python3 and PyQT5, supporting rectangular box, table, irregular text and key information annotation modes. Annotations can be directly used for the training of PP-OCR detection and recognition models.',
long_description=readme(),
long_description_content_type='text/markdown',
url='https://github.com/PaddlePaddle/PaddleOCR',
diff --git a/README.md b/README.md
index 8e869f6de5..7ad7d02860 100644
--- a/README.md
+++ b/README.md
@@ -26,12 +26,10 @@ PaddleOCR aims to create multilingual, awesome, leading, and practical OCR tools
## 📣 Recent updates
-- 💥 **Live Preview: Oct 24 - Oct 26, China Standard Time, 20:30**, Engineers@PaddleOCR will show PP-StructureV2 optimization strategy for 3 days.
- - Scan the QR code below using WeChat, follow the PaddlePaddle official account and fill out the questionnaire to join the WeChat group, get the live link and 20G OCR learning materials (including PDF2Word application, 10 models in vertical scenarios, etc.)
+- 🔨**2022.11 Add implementation of [4 cutting-edge algorithms](doc/doc_ch/algorithm_overview.md)**：Text Detection [DRRG](doc/doc_en/algorithm_det_drrg_en.md), Text Recognition [RFL](./doc/doc_en/algorithm_rec_rfl_en.md), Image Super-Resolution [Text Telescope](doc/doc_en/algorithm_sr_telescope_en.md)，Handwrittem Mathematical Expression Recognition [CAN](doc/doc_en/algorithm_rec_can_en.md)
+- **2022.10 release [optimized JS version PP-OCRv3 model](./deploy/paddlejs/README.md)** with 4.3M model size, 8x faster inference time, and a ready-to-use web demo
+- 💥 **Live Playback: Introduction to PP-StructureV2 optimization strategy**. Scan [the QR code below](#Community) using WeChat, follow the PaddlePaddle official account and fill out the questionnaire to join the WeChat group, get the live link and 20G OCR learning materials (including PDF2Word application, 10 models in vertical scenarios, etc.)
-
-
-
-
-
-
+
+
PaddleOCR官方交流群二维码
+
+
+
+
diff --git a/doc/doc_en/recognition_en.md b/doc/doc_en/recognition_en.md
index 7d31b0ffe2..78917aea90 100644
--- a/doc/doc_en/recognition_en.md
+++ b/doc/doc_en/recognition_en.md
@@ -15,6 +15,7 @@
* [2.6 Training with knowledge distillation](#kd)
* [2.7 Multi-language Training](#Multi_language)
* [2.8 Training on other platform(Windows/macOS/Linux DCU)](#28)
+ * [2.9 Fine-tuning](#29)
- [3. Evaluation and Test](#3-evaluation-and-test)
* [3.1 Evaluation](#31-evaluation)
* [3.2 Test](#32-test)
@@ -384,6 +385,11 @@ GPU mode is not supported, you need to set `use_gpu` to False in the configurati
- Linux DCU
Running on a DCU device requires setting the environment variable `export HIP_VISIBLE_DEVICES=0,1,2,3`, and the rest of the training and evaluation prediction commands are exactly the same as the Linux GPU.
+
+## 2.9 Fine-tuning
+
+In actual use, it is recommended to load the official pre-trained model and fine-tune it in your own data set. For the fine-tuning method of the recognition model, please refer to: [Model Fine-tuning Tutorial](./finetune_en.md).
+
## 3. Evaluation and Test
diff --git a/doc/doc_en/table_recognition_en.md b/doc/doc_en/table_recognition_en.md
index cff2933df2..d79d98936e 100644
--- a/doc/doc_en/table_recognition_en.md
+++ b/doc/doc_en/table_recognition_en.md
@@ -14,6 +14,9 @@ This article provides a full-process guide for the PaddleOCR table recognition m
- [2.5. Distributed Training](#25-distributed-training)
- [2.6. Training on other platform(Windows/macOS/Linux DCU)](#26-training-on-other-platformwindowsmacoslinux-dcu)
- [2.7. Fine-tuning](#27-fine-tuning)
+ - [2.7.1 Dataset](#271-dataset)
+ - [2.7.2 model selection](#272-model-selection)
+ - [2.7.3 Training hyperparameter selection](#273-training-hyperparameter-selection)
- [3. Evaluation and Test](#3-evaluation-and-test)
- [3.1. Evaluation](#31-evaluation)
- [3.2. Test table structure recognition effect](#32-test-table-structure-recognition-effect)
@@ -226,8 +229,40 @@ Running on a DCU device requires setting the environment variable `export HIP_VI
## 2.7. Fine-tuning
-In the actual use process, it is recommended to load the officially provided pre-training model and fine-tune it in your own data set. For the fine-tuning method of the table recognition model, please refer to: [Model fine-tuning tutorial](./finetune.md).
+### 2.7.1 Dataset
+
+Data number: It is recommended to prepare at least 2000 table recognition datasets for model fine-tuning.
+
+### 2.7.2 model selection
+
+It is recommended to choose the SLANet model (configuration file: [SLANet_ch.yml](../../configs/table/SLANet_ch.yml), pre-training model: [ch_ppstructure_mobile_v2.0_SLANet_train.tar](https://paddleocr.bj.bcebos .com/ppstructure/models/slanet/ch_ppstructure_mobile_v2.0_SLANet_train.tar)) for fine-tuning, its accuracy and generalization performance is the best Chinese table pre-training model currently available.
+
+For more table recognition models, please refer to [PP-Structure Series Model Library](../../ppstructure/docs/models_list.md).
+
+### 2.7.3 Training hyperparameter selection
+
+When fine-tuning the model, the most important hyperparameters are the pretrained model path `pretrained_model`, the learning rate `learning_rate`, and some configuration files are shown below.
+
+```yaml
+Global:
+ pretrained_model: ./ch_ppstructure_mobile_v2.0_SLANet_train/best_accuracy.pdparams # Pre-trained model path
+Optimizer:
+ lr:
+ name: Cosine
+ learning_rate: 0.001 #
+ warmup_epoch: 0
+ regularizer:
+ name: 'L2'
+ factor: 0
+```
+
+In the above configuration file, you first need to specify the `pretrained_model` field as the `best_accuracy.pdparams` file path.
+
+The configuration file provided by PaddleOCR is for 4-card training (equivalent to a total batch size of `4*48=192`) and no pre-trained model is loaded. Therefore, in your scenario, the learning rate is the same as the total The batch size needs to be adjusted linearly, for example
+
+* If your scenario is single card training, single card batch_size=48, then the total batch_size=48, it is recommended to adjust the learning rate to about `0.00025`.
+* If your scenario is for single-card training, due to memory limitations, you can only set batch_size=32 for a single card, then the total batch_size=32, it is recommended to adjust the learning rate to about `0.00017`.
# 3. Evaluation and Test
diff --git a/doc/doc_en/whl_en.md b/doc/doc_en/whl_en.md
index 77e80faa68..5628dc3f45 100644
--- a/doc/doc_en/whl_en.md
+++ b/doc/doc_en/whl_en.md
@@ -261,7 +261,7 @@ Output will be a list, each item contains classification result and confidence
## 3 Use custom model
When the built-in model cannot meet the needs, you need to use your own trained model.
-First, refer to the first section of [inference_en.md](./inference_en.md) to convert your det and rec model to inference model, and then use it as follows
+First, refer to [export](./detection_en.md#4-inference) doc to convert your det and rec model to inference model, and then use it as follows
### 3.1 Use by code
diff --git a/paddleocr.py b/paddleocr.py
index af0145b48b..96a641bb77 100644
--- a/paddleocr.py
+++ b/paddleocr.py
@@ -26,6 +26,9 @@
import logging
import numpy as np
from pathlib import Path
+import base64
+from io import BytesIO
+from PIL import Image
tools = importlib.import_module('.', 'tools')
ppocr = importlib.import_module('.', 'ppocr')
@@ -431,7 +434,25 @@ def check_img(img):
img, flag_gif, flag_pdf = check_and_read(image_file)
if not flag_gif and not flag_pdf:
with open(image_file, 'rb') as f:
- img = img_decode(f.read())
+ img_str = f.read()
+ img = img_decode(img_str)
+ if img is None:
+ try:
+ buf = BytesIO()
+ image = BytesIO(img_str)
+ im = Image.open(image)
+ rgb = im.convert('RGB')
+ rgb.save(buf, 'jpeg')
+ buf.seek(0)
+ image_bytes = buf.read()
+ data_base64 = str(base64.b64encode(image_bytes),
+ encoding="utf-8")
+ image_decode = base64.b64decode(data_base64)
+ img_array = np.frombuffer(image_decode, np.uint8)
+ img = cv2.imdecode(img_array, cv2.IMREAD_COLOR)
+ except:
+ logger.error("error in loading image:{}".format(image_file))
+ return None
if img is None:
logger.error("error in loading image:{}".format(image_file))
return None
diff --git a/ppocr/utils/e2e_metric/Deteval.py b/ppocr/utils/e2e_metric/Deteval.py
index 387b3c24e6..c2a4383eed 100755
--- a/ppocr/utils/e2e_metric/Deteval.py
+++ b/ppocr/utils/e2e_metric/Deteval.py
@@ -17,8 +17,6 @@
import scipy.io as io
from ppocr.utils.utility import check_install
-check_install("Polygon", "Polygon3")
-import Polygon as plg
from ppocr.utils.e2e_metric.polygon_fast import iod, area_of_intersection, area
@@ -279,6 +277,8 @@ def get_score_C(gt_label, text, pred_bboxes):
"""
get score for CentripetalText (CT) prediction.
"""
+ check_install("Polygon", "Polygon3")
+ import Polygon as plg
def gt_reading_mod(gt_label, text):
"""This helper reads groundtruths from mat files"""
diff --git a/ppstructure/kie/requirements.txt b/ppstructure/kie/requirements.txt
index 6cfcba7641..61c230d3ed 100644
--- a/ppstructure/kie/requirements.txt
+++ b/ppstructure/kie/requirements.txt
@@ -2,6 +2,6 @@ sentencepiece
yacs
seqeval
pypandoc
-attrdict
+attrdict3
python_docx
paddlenlp>=2.4.1
diff --git a/ppstructure/predict_system.py b/ppstructure/predict_system.py
index bb061c998f..b32b706299 100644
--- a/ppstructure/predict_system.py
+++ b/ppstructure/predict_system.py
@@ -229,7 +229,9 @@ def main(args):
if args.recovery and args.use_pdf2docx_api and flag_pdf:
from pdf2docx.converter import Converter
- docx_file = os.path.join(args.output, '{}.docx'.format(img_name))
+ os.makedirs(args.output, exist_ok=True)
+ docx_file = os.path.join(args.output,
+ '{}_api.docx'.format(img_name))
cv = Converter(image_file)
cv.convert(docx_file)
cv.close()
diff --git a/ppstructure/recovery/recovery_to_doc.py b/ppstructure/recovery/recovery_to_doc.py
index 1d8f8d9d4b..0501812082 100644
--- a/ppstructure/recovery/recovery_to_doc.py
+++ b/ppstructure/recovery/recovery_to_doc.py
@@ -73,7 +73,7 @@ def convert_info_docx(img, res, save_folder, img_name):
text_run.font.size = shared.Pt(10)
# save to docx
- docx_path = os.path.join(save_folder, '{}.docx'.format(img_name))
+ docx_path = os.path.join(save_folder, '{}_ocr.docx'.format(img_name))
doc.save(docx_path)
logger.info('docx save to {}'.format(docx_path))
diff --git a/ppstructure/recovery/requirements.txt b/ppstructure/recovery/requirements.txt
index ec08f9d0a2..761b9d7c3e 100644
--- a/ppstructure/recovery/requirements.txt
+++ b/ppstructure/recovery/requirements.txt
@@ -1,5 +1,4 @@
python-docx
-PyMuPDF==1.19.0
beautifulsoup4
fonttools>=4.24.0
fire>=0.3.0
diff --git a/ppstructure/recovery/table_process.py b/ppstructure/recovery/table_process.py
index 982e6b760f..77a6ef7659 100644
--- a/ppstructure/recovery/table_process.py
+++ b/ppstructure/recovery/table_process.py
@@ -278,8 +278,6 @@ def handle_table(self, html, doc):
cell_col += colspan
cell_row += 1
- doc.save('1.docx')
-
def handle_data(self, data):
if self.skip:
return
diff --git a/requirements.txt b/requirements.txt
index f3d9ce89e3..56f342a4e6 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -14,4 +14,4 @@ lxml
premailer
openpyxl
attrdict
-PyMuPDF==1.19.0
\ No newline at end of file
+PyMuPDF<1.21.0
diff --git a/test_tipc/test_serving_infer_cpp.sh b/test_tipc/test_serving_infer_cpp.sh
index 10ddecf3fa..6de685682a 100644
--- a/test_tipc/test_serving_infer_cpp.sh
+++ b/test_tipc/test_serving_infer_cpp.sh
@@ -103,7 +103,9 @@ function func_serving(){
last_status=${PIPESTATUS[0]}
eval "cat ${_save_log_path}"
status_check $last_status "${cpp_client_cmd}" "${status_log}" "${model_name}" "${_save_log_path}"
- ps ux | grep -i ${port_value} | awk '{print $2}' | xargs kill -s 9
+ #ps ux | grep -i ${port_value} | awk '{print $2}' | xargs kill -s 9
+ ${python_list[0]} ${web_service_py} stop
+ sleep 5s
else
server_log_path="${LOG_PATH}/cpp_server_gpu.log"
web_service_cpp_cmd="nohup ${python_list[0]} ${web_service_py} --model ${det_server_value} ${rec_server_value} ${op_key} ${op_value} ${port_key} ${port_value} ${gpu_key} ${gpu_id} > ${server_log_path} 2>&1 &"
@@ -115,7 +117,8 @@ function func_serving(){
last_status=${PIPESTATUS[0]}
eval "cat ${_save_log_path}"
status_check $last_status "${cpp_client_cmd}" "${status_log}" "${model_name}" "${_save_log_path}"
- ps ux | grep -i ${port_value} | awk '{print $2}' | xargs kill -s 9
+ #ps ux | grep -i ${port_value} | awk '{print $2}' | xargs kill -s 9
+ ${python_list[0]} ${web_service_py} stop
fi
done
}
diff --git a/tools/infer_kie_token_ser_re.py b/tools/infer_kie_token_ser_re.py
index c4fa2c927a..76120a913f 100755
--- a/tools/infer_kie_token_ser_re.py
+++ b/tools/infer_kie_token_ser_re.py
@@ -81,7 +81,7 @@ def make_input(ser_inputs, ser_results):
end.append(entity['end'])
label.append(entities_labels[res['pred']])
- entities = np.full([max_seq_len + 1, 3], fill_value=-1)
+ entities = np.full([max_seq_len + 1, 3], fill_value=-1, dtype=np.int64)
entities[0, 0] = len(start)
entities[1:len(start) + 1, 0] = start
entities[0, 1] = len(end)
@@ -98,7 +98,7 @@ def make_input(ser_inputs, ser_results):
head.append(i)
tail.append(j)
- relations = np.full([len(head) + 1, 2], fill_value=-1)
+ relations = np.full([len(head) + 1, 2], fill_value=-1, dtype=np.int64)
relations[0, 0] = len(head)
relations[1:len(head) + 1, 0] = head
relations[0, 1] = len(tail)