Merge branch 'dygraph' of https://github.com/PaddlePaddle/PaddleOCR into fix_vqa

.gitignore

@@ -29,3 +29,5 @@ paddleocr.egg-info/
 /deploy/android_demo/app/PaddleLite/
 /deploy/android_demo/app/.cxx/
 /deploy/android_demo/app/cache/
+test_tipc/web/models/
+test_tipc/web/node_modules/

PPOCRLabel/PPOCRLabel.py

@@ -61,7 +61,7 @@ from combobox import ComboBox
 from libs.constants import *
 from libs.utils import *
 from libs.settings import Settings
-from libs.shape import Shape, DEFAULT_LINE_COLOR, DEFAULT_FILL_COLOR
+from libs.shape import Shape, DEFAULT_LINE_COLOR, DEFAULT_FILL_COLOR,DEFAULT_LOCK_COLOR  
 from libs.stringBundle import StringBundle
 from libs.canvas import Canvas
 from libs.zoomWidget import ZoomWidget
@@ -126,7 +126,7 @@ class MainWindow(QMainWindow, WindowMixin):
         self.labelHist = []
         self.lastOpenDir = None
         self.result_dic = []
-
+        self.result_dic_locked = []
         self.changeFileFolder = False
         self.haveAutoReced = False
         self.labelFile = None
@@ -395,6 +395,7 @@ class MainWindow(QMainWindow, WindowMixin):
 
         delete = action(getStr('delBox'), self.deleteSelectedShape,
                         'backspace', 'delete', getStr('delBoxDetail'), enabled=False)
+        
         copy = action(getStr('dupBox'), self.copySelectedShape,
                       'Ctrl+C', 'copy', getStr('dupBoxDetail'),
                       enabled=False)
@@ -405,6 +406,7 @@ class MainWindow(QMainWindow, WindowMixin):
         showAll = action(getStr('showBox'), partial(self.togglePolygons, True),
                          'Ctrl+A', 'hide', getStr('showAllBoxDetail'),
                          enabled=False)
+            
 
         help = action(getStr('tutorial'), self.showTutorialDialog, None, 'help', getStr('tutorialDetail'))
         showInfo = action(getStr('info'), self.showInfoDialog, None, 'help', getStr('info'))
@@ -476,6 +478,10 @@ class MainWindow(QMainWindow, WindowMixin):
 
         undo = action(getStr("undo"), self.undoShapeEdit,
                       'Ctrl+Z', "undo", getStr("undo"), enabled=False)
+        
+        lock = action(getStr("lockBox"), self.lockSelectedShape,
+                      None, "lock", getStr("lockBoxDetail"),
+                      enabled=False)
 
         self.editButton.setDefaultAction(edit)
         self.newButton.setDefaultAction(create)
@@ -538,13 +544,13 @@ class MainWindow(QMainWindow, WindowMixin):
                               fitWindow=fitWindow, fitWidth=fitWidth,
                               zoomActions=zoomActions, saveLabel=saveLabel,
                               undo=undo, undoLastPoint=undoLastPoint,open_dataset_dir=open_dataset_dir,
-                              rotateLeft=rotateLeft,rotateRight=rotateRight,
+                              rotateLeft=rotateLeft,rotateRight=rotateRight,lock=lock,
                               fileMenuActions=(
                                   opendir,  open_dataset_dir, saveLabel,  resetAll, quit),
                               beginner=(), advanced=(),
                               editMenu=(createpoly, edit, copy, delete,singleRere,None, undo, undoLastPoint,
-                                        None, rotateLeft, rotateRight, None, color1, self.drawSquaresOption),
-                              beginnerContext=(create, edit, copy, delete, singleRere, rotateLeft, rotateRight,),
+                                        None, rotateLeft, rotateRight, None, color1, self.drawSquaresOption,lock),
+                              beginnerContext=(create, edit, copy, delete, singleRere, rotateLeft, rotateRight,lock),
                               advancedContext=(createMode, editMode, edit, copy,
                                                delete, shapeLineColor, shapeFillColor),
                               onLoadActive=(
@@ -998,6 +1004,7 @@ class MainWindow(QMainWindow, WindowMixin):
         self.actions.delete.setEnabled(n_selected)
         self.actions.copy.setEnabled(n_selected)
         self.actions.edit.setEnabled(n_selected == 1)
+        self.actions.lock.setEnabled(n_selected)
 
     def addLabel(self, shape):
         shape.paintLabel = self.displayLabelOption.isChecked()
@@ -1041,7 +1048,7 @@ class MainWindow(QMainWindow, WindowMixin):
     def loadLabels(self, shapes):
         s = []
         for label, points, line_color, fill_color, difficult in shapes:
-            shape = Shape(label=label)
+            shape = Shape(label=label,line_color=line_color)
             for x, y in points:
 
                 # Ensure the labels are within the bounds of the image. If not, fix them.
@@ -1051,6 +1058,7 @@ class MainWindow(QMainWindow, WindowMixin):
 
                 shape.addPoint(QPointF(x, y))
             shape.difficult = difficult
+            #shape.locked = False
             shape.close()
             s.append(shape)
 
@@ -1063,10 +1071,12 @@ class MainWindow(QMainWindow, WindowMixin):
             #     shape.fill_color = QColor(*fill_color)
             # else:
             #     shape.fill_color = generateColorByText(label)
-
+            
             self.addLabel(shape)
+ 
         self.updateComboBox()
         self.canvas.loadShapes(s)
+         
 
     def singleLabel(self, shape):
         if shape is None:
@@ -1106,10 +1116,9 @@ class MainWindow(QMainWindow, WindowMixin):
                         difficult=s.difficult)  # bool
 
         shapes = [] if mode == 'Auto' else \
-            [format_shape(shape) for shape in self.canvas.shapes]
+            [format_shape(shape) for shape in self.canvas.shapes if shape.line_color != DEFAULT_LOCK_COLOR]
         # Can add differrent annotation formats here
-
-        for box in self.result_dic:
+        for box in self.result_dic :
             trans_dic = {"label": box[1][0], "points": box[0], 'difficult': False}
             if trans_dic["label"] == "" and mode == 'Auto':
                 continue
@@ -1120,7 +1129,6 @@ class MainWindow(QMainWindow, WindowMixin):
             for box in shapes:
                 trans_dic.append({"transcription": box['label'], "points": box['points'], 'difficult': box['difficult']})
             self.PPlabel[annotationFilePath] = trans_dic
-
             if mode == 'Auto':
                 self.Cachelabel[annotationFilePath] = trans_dic
 
@@ -1313,6 +1321,7 @@ class MainWindow(QMainWindow, WindowMixin):
         # unicodeFilePath = os.path.abspath(unicodeFilePath)
         # Tzutalin 20160906 : Add file list and dock to move faster
         # Highlight the file item
+        
         if unicodeFilePath and self.fileListWidget.count() > 0:
             if unicodeFilePath in self.mImgList:
                 index = self.mImgList.index(unicodeFilePath)
@@ -1322,6 +1331,7 @@ class MainWindow(QMainWindow, WindowMixin):
                 ###
                 self.iconlist.clear()
                 self.additems5(None)
+             
                 for i in range(5):
                     item_tooltip = self.iconlist.item(i).toolTip()
                     # print(i,"---",item_tooltip)
@@ -1340,7 +1350,6 @@ class MainWindow(QMainWindow, WindowMixin):
 
         if unicodeFilePath and os.path.exists(unicodeFilePath):
             self.canvas.verified = False
-
             cvimg = cv2.imdecode(np.fromfile(unicodeFilePath, dtype=np.uint8), 1)
             height, width, depth = cvimg.shape
             cvimg = cv2.cvtColor(cvimg, cv2.COLOR_BGR2RGB)
@@ -1361,16 +1370,19 @@ class MainWindow(QMainWindow, WindowMixin):
             else:
                 self.dirty = False
                 self.actions.save.setEnabled(True)
-
+            if len(self.canvas.lockedShapes) != 0:
+                self.actions.save.setEnabled(True)
+                self.setDirty()
             self.canvas.setEnabled(True)
             self.adjustScale(initial=True)
             self.paintCanvas()
             self.addRecentFile(self.filePath)
             self.toggleActions(True)
+
             self.showBoundingBoxFromPPlabel(filePath)
 
             self.setWindowTitle(__appname__ + ' ' + filePath)
-
+            
             # Default : select last item if there is at least one item
             if self.labelList.count():
                 self.labelList.setCurrentItem(self.labelList.item(self.labelList.count() - 1))
@@ -1380,15 +1392,23 @@ class MainWindow(QMainWindow, WindowMixin):
             return True
         return False
 
-
     def showBoundingBoxFromPPlabel(self, filePath):
+        width, height = self.image.width(), self.image.height()
         imgidx = self.getImglabelidx(filePath)
-        if imgidx not in self.PPlabel.keys():
-            return
-        shapes = []
-        for box in self.PPlabel[imgidx]:
-            shapes.append((box['transcription'], box['points'], None, None, box['difficult']))
-
+        shapes =[]
+        #box['ratio'] of the shapes saved in lockedShapes contains the ratio of the
+        # four corner coordinates of the shapes to the height and width of the image
+        for box in self.canvas.lockedShapes:
+            if self.canvas.isInTheSameImage:
+                shapes.append((box['transcription'], [[s[0]*width,s[1]*height]for s in box['ratio']],
+                DEFAULT_LOCK_COLOR, None, box['difficult']))
+            else:
+                shapes.append(('锁定框：待检测', [[s[0]*width,s[1]*height]for s in box['ratio']],
+                DEFAULT_LOCK_COLOR, None, box['difficult']))
+        if imgidx in self.PPlabel.keys():
+            for box in self.PPlabel[imgidx]:
+                shapes.append((box['transcription'], box['points'], None, None, box['difficult']))
+ 
         self.loadLabels(shapes)
         self.canvas.verified = False
 
@@ -1646,9 +1666,37 @@ class MainWindow(QMainWindow, WindowMixin):
             else:
                 return fullFilePath
         return ''
-
+    
+   
+    def saveLockedShapes(self):
+        self.canvas.lockedShapes = []
+        self.canvas.selectedShapes = []
+        for s in self.canvas.shapes:
+            if s.line_color == DEFAULT_LOCK_COLOR:
+                self.canvas.selectedShapes.append(s)
+        self.lockSelectedShape()
+        for s in self.canvas.shapes:
+            if s.line_color == DEFAULT_LOCK_COLOR:
+                self.canvas.selectedShapes.remove(s)
+                self.canvas.shapes.remove(s)
+
+    
     def _saveFile(self, annotationFilePath, mode='Manual'):
+        if len(self.canvas.lockedShapes) != 0:
+            self.saveLockedShapes()
+
         if mode == 'Manual':
+            self.result_dic_locked = []
+            img = cv2.imread(self.filePath)
+            width, height = self.image.width(), self.image.height()
+            for shape in self.canvas.lockedShapes:
+                box = [[int(p[0]*width), int(p[1]*height)] for p in shape['ratio']]
+                assert len(box) == 4
+                result = [(shape['transcription'],1)]
+                result.insert(0, box)
+                self.result_dic_locked.append(result)
+            self.result_dic += self.result_dic_locked
+            self.result_dic_locked = []
             if annotationFilePath and self.saveLabels(annotationFilePath, mode=mode):
                 self.setClean()
                 self.statusBar().showMessage('Saved to  %s' % annotationFilePath)
@@ -1663,13 +1711,13 @@ class MainWindow(QMainWindow, WindowMixin):
                     self.savePPlabel(mode='Auto')
 
                 self.fileListWidget.insertItem(int(currIndex), item)
-                self.openNextImg()
+                if not self.canvas.isInTheSameImage:
+                    self.openNextImg()
                 self.actions.saveRec.setEnabled(True)
                 self.actions.saveLabel.setEnabled(True)
 
         elif mode == 'Auto':
             if annotationFilePath and self.saveLabels(annotationFilePath, mode=mode):
-
                 self.setClean()
                 self.statusBar().showMessage('Saved to  %s' % annotationFilePath)
                 self.statusBar().show()
@@ -1733,7 +1781,9 @@ class MainWindow(QMainWindow, WindowMixin):
             if discardChanges == QMessageBox.No:
                 return True
             elif discardChanges == QMessageBox.Yes:
+                self.canvas.isInTheSameImage = True
                 self.saveFile()
+                self.canvas.isInTheSameImage = False
                 return True
             else:
                 return False
@@ -1872,6 +1922,7 @@ class MainWindow(QMainWindow, WindowMixin):
         # org_box = [dic['points'] for dic in self.PPlabel[self.getImglabelidx(self.filePath)]]
         if self.canvas.shapes:
             self.result_dic = []
+            self.result_dic_locked = []  # result_dic_locked stores the ocr result of self.canvas.lockedShapes
             rec_flag = 0
             for shape in self.canvas.shapes:
                 box = [[int(p.x()), int(p.y())] for p in shape.points]
@@ -1883,21 +1934,32 @@ class MainWindow(QMainWindow, WindowMixin):
                     return
                 result = self.ocr.ocr(img_crop, cls=True, det=False)
                 if result[0][0] != '':
-                    result.insert(0, box)
-                    print('result in reRec is ', result)
-                    self.result_dic.append(result)
+                    if shape.line_color == DEFAULT_LOCK_COLOR:
+                        shape.label = result[0][0]
+                        result.insert(0, box)
+                        self.result_dic_locked.append(result)
+                    else:
+                        result.insert(0, box)
+                        self.result_dic.append(result)
                 else:
                     print('Can not recognise the box')
-                    self.result_dic.append([box,(self.noLabelText,0)])
-
-                if self.noLabelText == shape.label or result[1][0] == shape.label:
-                    print('label no change')
-                else:
-                    rec_flag += 1
-
-            if len(self.result_dic) > 0 and rec_flag > 0:
+                    if shape.line_color == DEFAULT_LOCK_COLOR:
+                        shape.label = result[0][0]
+                        self.result_dic_locked.append([box,(self.noLabelText,0)])
+                    else:
+                        self.result_dic.append([box,(self.noLabelText,0)])
+                try:
+                    if self.noLabelText == shape.label or result[1][0] == shape.label:
+                        print('label no change')
+                    else:
+                        rec_flag += 1
+                except IndexError as e:
+                    print('Can not recognise the box')           
+            if (len(self.result_dic) > 0 and rec_flag > 0)or self.canvas.lockedShapes: 
+                self.canvas.isInTheSameImage = True              
                 self.saveFile(mode='Auto')
                 self.loadFile(self.filePath)
+                self.canvas.isInTheSameImage = False
                 self.setDirty()
             elif len(self.result_dic) == len(self.canvas.shapes) and rec_flag == 0:
                 QMessageBox.information(self, "Information", "The recognition result remains unchanged!")
@@ -2107,6 +2169,44 @@ class MainWindow(QMainWindow, WindowMixin):
         self.labelList.clearSelection()
         self._noSelectionSlot = False
         self.canvas.loadShapes(shapes, replace=replace)
+        print("loadShapes")#1
+        
+    
+    def lockSelectedShape(self):
+        """lock the selsected shapes.
+
+        Add self.selectedShapes to lock self.canvas.lockedShapes, 
+        which holds the ratio of the four coordinates of the locked shapes
+        to the width and height of the image
+        """
+        width, height = self.image.width(), self.image.height()
+        def format_shape(s):
+            return dict(label=s.label,  # str
+                        line_color=s.line_color.getRgb(),
+                        fill_color=s.fill_color.getRgb(),
+                        ratio=[[int(p.x())/width, int(p.y())/height] for p in s.points],  # QPonitF
+                       # add chris
+                        difficult=s.difficult)  # bool
+        #lock
+        if len(self.canvas.lockedShapes) == 0:
+            for s in self.canvas.selectedShapes:
+                s.line_color = DEFAULT_LOCK_COLOR
+                s.locked = True
+            shapes = [format_shape(shape) for shape in self.canvas.selectedShapes]
+            trans_dic = []
+            for box in shapes:
+                trans_dic.append({"transcription": box['label'], "ratio": box['ratio'], 'difficult': box['difficult']})
+            self.canvas.lockedShapes = trans_dic
+            self.actions.save.setEnabled(True)
+
+        #unlock
+        else:
+            for s in self.canvas.shapes:
+                s.line_color = DEFAULT_LINE_COLOR
+            self.canvas.lockedShapes = []
+            self.result_dic_locked = []
+            self.setDirty()
+            self.actions.save.setEnabled(True)
 
 
 def inverted(color):

PPOCRLabel/README.md

@@ -78,14 +78,14 @@ PPOCRLabel # run
 ```bash
 cd PaddleOCR/PPOCRLabel
 python3 setup.py bdist_wheel 
-pip3 install dist/PPOCRLabel-1.0.0-py2.py3-none-any.whl
+pip3 install dist/PPOCRLabel-1.0.2-py2.py3-none-any.whl
 ```
 
 #### 1.2.3 Run PPOCRLabel by Python Script
 
 ```bash
 cd ./PPOCRLabel  # Switch to the PPOCRLabel directory
-python PPOCRLabel.py --lang ch
+python PPOCRLabel.py
 ```
 
 

PPOCRLabel/README_ch.md

@@ -78,7 +78,7 @@ PPOCRLabel --lang ch # 启动
 ```bash
 cd PaddleOCR/PPOCRLabel
 python3 setup.py bdist_wheel 
-pip3 install dist/PPOCRLabel-1.0.0-py2.py3-none-any.whl -i https://mirror.baidu.com/pypi/simple
+pip3 install dist/PPOCRLabel-1.0.2-py2.py3-none-any.whl -i https://mirror.baidu.com/pypi/simple
 ```
 
 #### 1.2.3 通过Python脚本运行PPOCRLabel

PPOCRLabel/libs/canvas.py

@@ -87,6 +87,10 @@ class Canvas(QWidget):
         #initialisation for panning
         self.pan_initial_pos = QPoint()
 
+        #lockedshapes related 
+        self.lockedShapes = []
+        self.isInTheSameImage = False
+
     def setDrawingColor(self, qColor):
         self.drawingLineColor = qColor
         self.drawingRectColor = qColor

PPOCRLabel/libs/shape.py

@@ -30,6 +30,7 @@ DEFAULT_SELECT_LINE_COLOR = QColor(255, 255, 255)
 DEFAULT_SELECT_FILL_COLOR = QColor(0, 128, 255, 155)
 DEFAULT_VERTEX_FILL_COLOR = QColor(0, 255, 0, 255)
 DEFAULT_HVERTEX_FILL_COLOR = QColor(255, 0, 0)
+DEFAULT_LOCK_COLOR = QColor(255, 0, 255)
 MIN_Y_LABEL = 10
 
 
@@ -57,7 +58,7 @@ class Shape(object):
         self.selected = False
         self.difficult = difficult
         self.paintLabel = paintLabel
-
+        self.locked = False
         self._highlightIndex = None
         self._highlightMode = self.NEAR_VERTEX
         self._highlightSettings = {

PPOCRLabel/libs/stringBundle.py

@@ -60,7 +60,7 @@ class StringBundle:
 
     def __createLookupFallbackList(self, localeStr):
         resultPaths = []
-        basePath = "\strings" if os.name == 'nt' else ":/strings"
+        basePath = "\strings" if os.name == 'nt' else "/strings"
         resultPaths.append(basePath)
         if localeStr is not None:
             # Don't follow standard BCP47. Simple fallback

PPOCRLabel/resources/strings/strings-en.properties

@@ -104,4 +104,6 @@ singleRe=Re-recognition RectBox
 labelDialogOption=Pop-up Label Input Dialog
 undo=Undo
 undoLastPoint=Undo Last Point
-autoSaveMode=Auto Export Label Mode
\ No newline at end of file
+autoSaveMode=Auto Export Label Mode
+lockBox=Lock selected box/Unlock all box
+lockBoxDetail=Lock selected box/Unlock all box
\ No newline at end of file

PPOCRLabel/resources/strings/strings-zh-CN.properties

@@ -104,4 +104,6 @@ singleRe=重识别此区块
 labelDialogOption=弹出标记输入框
 undo=撤销
 undoLastPoint=撤销上个点
-autoSaveMode=自动导出标记结果
\ No newline at end of file
+autoSaveMode=自动导出标记结果
+lockBox=锁定框/解除锁定框
+lockBoxDetail=若当前没有框处于锁定状态则锁定选中的框，若存在锁定框则解除所有锁定框的锁定状态

PPOCRLabel/setup.py

@@ -33,7 +33,7 @@ setup(
     package_dir={'PPOCRLabel': ''},
     include_package_data=True,
     entry_points={"console_scripts": ["PPOCRLabel= PPOCRLabel.PPOCRLabel:main"]},
-    version='1.0.0',
+    version='1.0.2',
     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',

README.md

@@ -24,7 +24,7 @@ PaddleOCR aims to create multilingual, awesome, leading, and practical OCR tools
 
 **Recent updates**
 - 2021.12.21 OCR open source online course starts. The lesson starts at 8:30 every night and lasts for ten days. Free registration: https://aistudio.baidu.com/aistudio/course/introduce/25207
-- 2021.12.21 release PaddleOCR v2.4, release 1 text detection algorithm (PSENet), 3 text recognition algorithms (NRTR、SEED、SAR), 1 key information extraction algorithm (SDMGR) and 3 DocVQA algorithms (LayoutLM、LayoutLMv2，LayoutXLM).
+- 2021.12.21 release PaddleOCR v2.4, release 1 text detection algorithm (PSENet), 3 text recognition algorithms (NRTR、SEED、SAR), 1 key information extraction algorithm (SDMGR, [tutorial](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.4/ppstructure/docs/kie.md)) and 3 DocVQA algorithms (LayoutLM, LayoutLMv2, LayoutXLM, [tutorial](https://github.com/PaddlePaddle/PaddleOCR/tree/release/2.4/ppstructure/vqa)).
 - PaddleOCR R&D team would like to share the key points of PP-OCRv2, at 20:15 pm on September 8th, [Course Address](https://aistudio.baidu.com/aistudio/education/group/info/6758).
 - 2021.9.7 release PaddleOCR v2.3, [PP-OCRv2](#PP-OCRv2) is proposed. The inference speed of PP-OCRv2 is 220% higher than that of PP-OCR server in CPU device. The F-score of PP-OCRv2 is 7% higher than that of PP-OCR mobile.
 - 2021.8.3 released PaddleOCR v2.2, add a new structured documents analysis toolkit, i.e., [PP-Structure](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/ppstructure/README.md), support layout analysis and table recognition (One-key to export chart images to Excel files).
@@ -39,7 +39,7 @@ PaddleOCR aims to create multilingual, awesome, leading, and practical OCR tools
     - General PP-OCR server series models: detection (47.1M) + direction classifier (1.4M) + recognition (94.9M) = 143.4M
     - Support Chinese, English, and digit recognition, vertical text recognition, and long text recognition
     - Support multi-language recognition: about 80 languages like Korean, Japanese, German, French, etc
-- document structurize system PP-Structure
+- PP-Structure: a document structurize system
     - support layout analysis and table recognition (support export to Excel)
     - support key information extraction
     - support DocVQA
@@ -90,7 +90,7 @@ Mobile DEMO experience (based on EasyEdge and Paddle-Lite, supports iOS and Andr
 
 | Model introduction                                           | Model name                   | Recommended scene | Detection model                                              | Direction classifier                                         | Recognition model                                            |
 | ------------------------------------------------------------ | ---------------------------- | ----------------- | ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ |
-| Chinese and English ultra-lightweight PP-OCRv2 model（11.6M） |  ch_PP-OCRv2_xx |Mobile & Server|[inference model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_distill_train.tar)| [inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[inference model](https://paddleocr.bj.bcebos.com/PP-OCRv2/ch/ch_PP-OCRv2_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_train.tar)|
+| Chinese and English ultra-lightweight PP-OCRv2 model（11.6M） |  ch_PP-OCRv2_xx |Mobile & Server|[inference model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_distill_train.tar)| [inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[inference model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_train.tar)|
 | Chinese and English ultra-lightweight PP-OCR model (9.4M)       | ch_ppocr_mobile_v2.0_xx      | Mobile & server   |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_train.tar)|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_train.tar)      |
 | Chinese and English general PP-OCR model (143.4M)               | ch_ppocr_server_v2.0_xx      | Server            |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_train.tar)    |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_traingit.tar)    |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_train.tar)  |
 
@@ -102,11 +102,11 @@ For a new language request, please refer to [Guideline for new language_requests
 ## Tutorials
 - [Environment Preparation](./doc/doc_en/environment_en.md)
 - [Quick Start](./doc/doc_en/quickstart_en.md)
-- [PaddleOCR Overview and Installation](./doc/doc_en/paddleOCR_overview_en.md)
+- [PaddleOCR Overview and Project Clone](./doc/doc_en/paddleOCR_overview_en.md)
 - PP-OCR Industry Landing: from Training to Deployment
-    - [PP-OCR Model and Configuration](./doc/doc_en/models_and_config_en.md)
+    - [PP-OCR Model Zoo](./doc/doc_en/models_en.md)
         - [PP-OCR Model Download](./doc/doc_en/models_list_en.md)
-        - [Python Inference for PP-OCR Model Library](./doc/doc_en/inference_ppocr_en.md)
+        - [Python Inference for PP-OCR Model Zoo](./doc/doc_en/inference_ppocr_en.md)
     - [PP-OCR Training](./doc/doc_en/training_en.md)
         - [Text Detection](./doc/doc_en/detection_en.md)
         - [Text Recognition](./doc/doc_en/recognition_en.md)
@@ -120,6 +120,8 @@ For a new language request, please refer to [Guideline for new language_requests
 - [PP-Structure: Information Extraction](./ppstructure/README.md)
     - [Layout Parser](./ppstructure/layout/README.md)
     - [Table Recognition](./ppstructure/table/README.md)
+    - [DocVQA](https://github.com/PaddlePaddle/PaddleOCR/tree/release/2.4/ppstructure/vqa)
+    - [Key Information Extraction](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.4/ppstructure/docs/kie.md)
 - Academic Circles
     - [Two-stage Algorithm](./doc/doc_en/algorithm_overview_en.md)
     - [PGNet Algorithm](./doc/doc_en/pgnet_en.md)

README_ch.md

@@ -19,8 +19,8 @@ PaddleOCR旨在打造一套丰富、领先、且实用的OCR工具库，助力
 
 ## 近期更新
 
-- 2021.12.21 《OCR十讲》课程开讲，12月21日起每晚八点半线上授课！ 【免费】报名地址：https://aistudio.baidu.com/aistudio/course/introduce/25207
-- 2021.12.21 发布PaddleOCR v2.4。OCR算法新增1种文本检测算法（PSENet），3种文本识别算法（NRTR、SEED、SAR）；文档结构化算法新增1种关键信息提取算法（SDMGR），3种DocVQA算法（LayoutLM、LayoutLMv2，LayoutXLM）。
+- 2021.12.21《动手学OCR · 十讲》课程开讲，12月21日起每晚八点半线上授课！[免费报名地址](https://aistudio.baidu.com/aistudio/course/introduce/25207)。
+- 2021.12.21 发布PaddleOCR v2.4。OCR算法新增1种文本检测算法（PSENet），3种文本识别算法（NRTR、SEED、SAR）；文档结构化算法新增1种关键信息提取算法（SDMGR，[文档](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.4/ppstructure/docs/kie.md)），3种DocVQA算法（LayoutLM、LayoutLMv2，LayoutXLM，[文档](https://github.com/PaddlePaddle/PaddleOCR/tree/release/2.4/ppstructure/vqa)）。
 - PaddleOCR研发团队对最新发版内容技术深入解读，9月8日晚上20:15，[课程回放](https://aistudio.baidu.com/aistudio/education/group/info/6758)。
 - 2021.9.7 发布PaddleOCR v2.3与[PP-OCRv2](#PP-OCRv2)，CPU推理速度相比于PP-OCR server提升220%；效果相比于PP-OCR mobile 提升7%。
 - 2021.8.3 发布PaddleOCR v2.2，新增文档结构分析[PP-Structure](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/ppstructure/README_ch.md)工具包，支持版面分析与表格识别（含Excel导出）。
@@ -54,8 +54,7 @@ PaddleOCR旨在打造一套丰富、领先、且实用的OCR工具库，助力
 
 - 加入社区：微信扫描下方二维码加入官方交流群，与各行各业开发者充分交流，期待您的加入。
 - 社区贡献：[社区贡献](./doc/doc_ch/thirdparty.md)文档中包含了社区用户**使用PaddleOCR开发的各种工具、应用**以及**为PaddleOCR贡献的功能、优化的文档与代码**等，是官方为社区开发者打造的荣誉墙、也是帮助优质项目宣传的广播站。如果您的OCR项目未被收集在文档中，可根据文档说明与我们联系。最新社区贡献可查看[此处](#社区贡献)。
-
-- 社区常规赛：作为社区贡献的具体承载形式，社区常规赛是面向OCR开发者的积分赛事。首届社区常规赛与《动手学OCR · 十讲》课程联合推广，课程详情可参考[链接](https://aistudio.baidu.com/aistudio/course/introduce/25207)，课程奖励与作业说明可参考[链接](https://github.com/PaddlePaddle/PaddleOCR/issues/4982)。
+- 社区常规赛：作为社区贡献的具体承载形式，社区常规赛是面向OCR开发者的积分赛事。首届社区常规赛与[《动手学OCR · 十讲》课程](https://aistudio.baidu.com/aistudio/course/introduce/25207)联合推广。社区常规赛的赛题详情与报名方法可参考[链接](https://github.com/PaddlePaddle/PaddleOCR/issues/4982)。
 
 <div align="center">
 <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleOCR/dygraph/doc/joinus.PNG"  width = "200" height = "200" />
@@ -64,22 +63,33 @@ PaddleOCR旨在打造一套丰富、领先、且实用的OCR工具库，助力
 ## 零代码体验
 
 - 在线网站体验：超轻量PP-OCR mobile模型体验地址：https://www.paddlepaddle.org.cn/hub/scene/ocr
-
 - 移动端：[安装包DEMO下载地址](https://ai.baidu.com/easyedge/app/openSource?from=paddlelite)(基于EasyEdge和Paddle-Lite, 支持iOS和Android系统)
 
+<a name="模型下载"></a>
+## PP-OCR系列模型列表（更新中）
+
+| 模型简介                              | 模型名称                | 推荐场景        | 检测模型                                                     | 方向分类器                                                   | 识别模型                                                     |
+| ------------------------------------- | ----------------------- | --------------- | ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ |
+| 中英文超轻量PP-OCRv2模型（13.0M）     | ch_PP-OCRv2_xx          | 移动端&服务器端 | [推理模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_distill_train.tar) | [推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) | [推理模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_train.tar) |
+| 中英文超轻量PP-OCR mobile模型（9.4M） | ch_ppocr_mobile_v2.0_xx | 移动端&服务器端 | [推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_train.tar) | [推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) | [推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_pre.tar) |
+| 中英文通用PP-OCR server模型（143.4M） | ch_ppocr_server_v2.0_xx | 服务器端        | [推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_train.tar) | [推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) | [推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_pre.tar) |
+
+更多模型下载（包括多语言），可以参考[PP-OCR 系列模型下载](./doc/doc_ch/models_list.md)
 
 ## 文档教程
+
 - [运行环境准备](./doc/doc_ch/environment.md)
 - [快速开始（中英文/多语言/文档分析）](./doc/doc_ch/quickstart.md)
 - [PaddleOCR全景图与项目克隆](./doc/doc_ch/paddleOCR_overview.md)
 - PP-OCR产业落地：从训练到部署
-    - [PP-OCR模型与配置文件](./doc/doc_ch/models_and_config.md)
+    - [PP-OCR模型库](./doc/doc_ch/models.md)
         - [PP-OCR模型下载](./doc/doc_ch/models_list.md)
         - [PP-OCR模型库快速推理](./doc/doc_ch/inference_ppocr.md)
     - [PP-OCR模型训练](./doc/doc_ch/training.md)
         - [文本检测](./doc/doc_ch/detection.md)
         - [文本识别](./doc/doc_ch/recognition.md)
         - [文本方向分类器](./doc/doc_ch/angle_class.md)
+        - [知识蒸馏](./doc/doc_ch/knowledge_distillation.md)
         - [配置文件内容与生成](./doc/doc_ch/config.md)
     - PP-OCR模型推理部署
         - [基于C++预测引擎推理](./deploy/cpp_infer/readme.md)
@@ -89,6 +99,8 @@ PaddleOCR旨在打造一套丰富、领先、且实用的OCR工具库，助力
 - [PP-Structure信息提取](./ppstructure/README_ch.md)
     - [版面分析](./ppstructure/layout/README_ch.md)
     - [表格识别](./ppstructure/table/README_ch.md)
+    - [DocVQA](https://github.com/PaddlePaddle/PaddleOCR/tree/release/2.4/ppstructure/vqa)
+    - [关键信息提取](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.4/ppstructure/docs/kie.md)
 - OCR学术圈
     - [两阶段模型介绍与下载](./doc/doc_ch/algorithm_overview.md)
     - [端到端PGNet算法](./doc/doc_ch/pgnet.md)
@@ -119,7 +131,7 @@ PaddleOCR旨在打造一套丰富、领先、且实用的OCR工具库，助力
 </div>
 [1] PP-OCR是一个实用的超轻量OCR系统。主要由DB文本检测、检测框矫正和CRNN文本识别三部分组成。该系统从骨干网络选择和调整、预测头部的设计、数据增强、学习率变换策略、正则化参数选择、预训练模型使用以及模型自动裁剪量化8个方面，采用19个有效策略，对各个模块的模型进行效果调优和瘦身(如绿框所示)，最终得到整体大小为3.5M的超轻量中英文OCR和2.8M的英文数字OCR。更多细节请参考PP-OCR技术方案 https://arxiv.org/abs/2009.09941
 
-[2] PP-OCRv2在PP-OCR的基础上，进一步在5个方面重点优化，检测模型采用CML协同互学习知识蒸馏策略和CopyPaste数据增广策略；识别模型采用LCNet轻量级骨干网络、UDML 改进知识蒸馏策略和Enhanced CTC loss损失函数改进（如上图红框所示），进一步在推理速度和预测效果上取得明显提升。更多细节请参考PP-OCRv2[技术报告](https://arxiv.org/abs/2109.03144)。
+[2] PP-OCRv2在PP-OCR的基础上，进一步在5个方面重点优化，检测模型采用CML协同互学习知识蒸馏策略和CopyPaste数据增广策略；识别模型采用LCNet轻量级骨干网络、UDML 改进知识蒸馏策略和[Enhanced CTC loss](./doc/doc_ch/enhanced_ctc_loss.md)损失函数改进（如上图红框所示），进一步在推理速度和预测效果上取得明显提升。更多细节请参考PP-OCRv2[技术报告](https://arxiv.org/abs/2109.03144)。
 
 <a name="效果展示"></a>
 

configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_cml.yml

@@ -21,6 +21,7 @@ Architecture:
   model_type: det
   Models:
     Teacher:
+      pretrained: ./pretrain_models/ch_ppocr_server_v2.0_det_train/best_accuracy
       freeze_params: true
       return_all_feats: false
       model_type: det
@@ -36,6 +37,7 @@ Architecture:
         name: DBHead
         k: 50
     Student:
+      pretrained:
       freeze_params: false
       return_all_feats: false
       model_type: det
@@ -52,6 +54,7 @@ Architecture:
         name: DBHead
         k: 50
     Student2:
+      pretrained:
       freeze_params: false
       return_all_feats: false
       model_type: det

configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_distill.yml

@@ -18,6 +18,7 @@ Global:
 Architecture:
   name: DistillationModel
   algorithm: Distillation
+  model_type: det
   Models:
     Student:
       pretrained: ./pretrain_models/MobileNetV3_large_x0_5_pretrained

configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_dml.yml

@@ -18,6 +18,7 @@ Global:
 Architecture:
   name: DistillationModel
   algorithm: Distillation
+  model_type: det
   Models:
     Student:
       pretrained: ./pretrain_models/MobileNetV3_large_x0_5_pretrained

deploy/paddle2onnx/readme.md

@@ -18,8 +18,8 @@ python3.7 -m pip install paddle2onnx
 
 - 安装 ONNX
 ```
-# 建议安装 1.4.0 版本，可根据环境更换版本号
-python3.7 -m pip install onnxruntime==1.4.0
+# 建议安装 1.9.0 版本，可根据环境更换版本号
+python3.7 -m pip install onnxruntime==1.9.0
 ```
 
 ## 2. 模型转换
@@ -47,13 +47,15 @@ paddle2onnx --model_dir=./inference/ch_ppocr_mobile_v2.0_det_infer/ \
 --params_filename=inference.pdiparams \
 --save_file=./inference/det_mobile_onnx/model.onnx \
 --opset_version=10 \
+--input_shape_dict="{'x': [-1, 3, -1, -1]}" \
 --enable_onnx_checker=True
 ```
 
 执行完毕后，ONNX 模型会被保存在 `./inference/det_mobile_onnx/` 路径下
 
-* 注意：以下几个模型暂不支持转换为 ONNX 模型：
-NRTR、SAR、RARE、SRN
+* 注意：对于OCR模型，转化过程中必须采用动态shape的形式，即加入选项--input_shape_dict="{'x': [-1, 3, -1, -1]}"，否则预测结果可能与直接使用Paddle预测有细微不同。
+  另外，以下几个模型暂不支持转换为 ONNX 模型：
+  NRTR、SAR、RARE、SRN
 
 ## 3. onnx 预测
 
@@ -72,5 +74,3 @@ root INFO: 1.jpg  [[[291, 295], [334, 292], [348, 844], [305, 847]], [[344, 296]
 The predict time of ../../doc/imgs/1.jpg: 0.06162881851196289
 The visualized image saved in ./inference_results/det_res_1.jpg
 ```
-
-* 注意：ONNX暂时不支持变长预测，需要将输入resize到固定输入，预测结果可能与直接使用Paddle预测有细微不同。

doc/doc_ch/algorithm_overview.md

@@ -57,7 +57,7 @@ PaddleOCR基于动态图开源的文本识别算法列表：
 - [x]  SAR([paper](https://arxiv.org/abs/1811.00751v2))
 - [x] SEED([paper](https://arxiv.org/pdf/2005.10977.pdf))
 
-参考[DTRB][3](https://arxiv.org/abs/1904.01906)文字识别训练和评估流程，使用MJSynth和SynthText两个文字识别数据集训练，在IIIT, SVT, IC03, IC13, IC15, SVTP, CUTE数据集上进行评估，算法效果如下：
+参考[DTRB](https://arxiv.org/abs/1904.01906)[3]文字识别训练和评估流程，使用MJSynth和SynthText两个文字识别数据集训练，在IIIT, SVT, IC03, IC13, IC15, SVTP, CUTE数据集上进行评估，算法效果如下：
 
 |模型|骨干网络|Avg Accuracy|模型存储命名|下载链接|
 |---|---|---|---|---|

doc/doc_ch/code_and_doc.md

@@ -146,7 +146,7 @@ PaddleOCR欢迎大家向repo中积极贡献代码，下面给出一些贡献代
 - 将 `远程仓库` Clone到本地
 
 ```
-# 拉取develop分支的代码
+# 拉取dygraph分支的代码
 git clone https://github.com/{your_name}/PaddleOCR.git -b dygraph
 cd PaddleOCR
 ```
@@ -191,11 +191,11 @@ git checkout -b new_branch
 也可以基于远程或者上游的分支创建新的分支，命令如下。
 
 ```
-# 基于用户远程仓库(origin)的develop创建new_branch分支
-git checkout -b new_branch origin/develop
-# 基于上游远程仓库(upstream)的develop创建new_branch分支
+# 基于用户远程仓库(origin)的dygraph创建new_branch分支
+git checkout -b new_branch origin/dygraph
+# 基于上游远程仓库(upstream)的dygraph创建new_branch分支
 # 如果需要从upstream创建新的分支，需要首先使用git fetch upstream获取上游代码
-git checkout -b new_branch upstream/develop
+git checkout -b new_branch upstream/dygraph
 ```
 
 最终会显示切换到新的分支，输出信息如下
@@ -246,8 +246,8 @@ git commit -m "your commit info"
 
 ```
 git fetch upstream
-# 如果是希望提交到其他分支，则需要从upstream的其他分支pull代码，这里是develop
-git pull upstream develop
+# 如果是希望提交到其他分支，则需要从upstream的其他分支pull代码，这里是dygraph
+git pull upstream dygraph
 ```
 
 #### 3.2.7 push到远程仓库