Skip to content

P
PaddleOCR

s920243400 / PaddleOCR
与 Fork 源项目一致

Fork自 PaddlePaddle / PaddleOCR

通知 1
Star 1
Fork 0
P
PaddleOCR
You need to sign in or sign up before continuing.
提交 6ec9aa4d 编写于 作者: L LDOUBLEV
浏览文件
操作

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

上级 df1c97af 65e61f44
显示空白变更内容
内联 并排
Showing with 11352 addition and 11109 deletion
PPOCRLabel/PPOCRLabel.py
浏览文件 @ 6ec9aa4d
......@@ -206,7 +206,7 @@ class MainWindow(QMainWindow, WindowMixin):
self.labelList = EditInList()
labelListContainer = QWidget()
labelListContainer.setLayout(listLayout)
self.labelList.itemActivated.connect(self.labelSelectionChanged)
#self.labelList.itemActivated.connect(self.labelSelectionChanged)
self.labelList.itemSelectionChanged.connect(self.labelSelectionChanged)
self.labelList.clicked.connect(self.labelList.item_clicked)
# Connect to itemChanged to detect checkbox changes.
......@@ -219,7 +219,7 @@ class MainWindow(QMainWindow, WindowMixin):
################## detection box ####################
self.BoxList = QListWidget()
self.BoxList.itemActivated.connect(self.boxSelectionChanged)
#self.BoxList.itemActivated.connect(self.boxSelectionChanged)
self.BoxList.itemSelectionChanged.connect(self.boxSelectionChanged)
self.BoxList.itemDoubleClicked.connect(self.editBox)
# Connect to itemChanged to detect checkbox changes.
......@@ -435,7 +435,7 @@ class MainWindow(QMainWindow, WindowMixin):
######## New actions #######
AutoRec = action(getStr('autoRecognition'), self.autoRecognition,
'Ctrl+Shift+A', 'Auto', getStr('autoRecognition'), enabled=False)
'', 'Auto', getStr('autoRecognition'), enabled=False)
reRec = action(getStr('reRecognition'), self.reRecognition,
'Ctrl+Shift+R', 'reRec', getStr('reRecognition'), enabled=False)
......@@ -444,7 +444,7 @@ class MainWindow(QMainWindow, WindowMixin):
'Ctrl+R', 'reRec', getStr('singleRe'), enabled=False)
createpoly = action(getStr('creatPolygon'), self.createPolygon,
'q', 'new', 'Creat Polygon', enabled=True)
'q', 'new', getStr('creatPolygon'), enabled=True)
saveRec = action(getStr('saveRec'), self.saveRecResult,
'', 'save', getStr('saveRec'), enabled=False)
......@@ -452,6 +452,12 @@ class MainWindow(QMainWindow, WindowMixin):
saveLabel = action(getStr('saveLabel'), self.saveLabelFile, #
'Ctrl+S', 'save', getStr('saveLabel'), enabled=False)
undoLastPoint = action(getStr("undoLastPoint"), self.canvas.undoLastPoint,
'Ctrl+Z', "undo", getStr("undoLastPoint"), enabled=False)
undo = action(getStr("undo"), self.undoShapeEdit,
'Ctrl+Z', "undo", getStr("undo"), enabled=False)
self.editButton.setDefaultAction(edit)
self.newButton.setDefaultAction(create)
self.DelButton.setDefaultAction(deleteImg)
......@@ -512,10 +518,11 @@ class MainWindow(QMainWindow, WindowMixin):
zoom=zoom, zoomIn=zoomIn, zoomOut=zoomOut, zoomOrg=zoomOrg,
fitWindow=fitWindow, fitWidth=fitWidth,
zoomActions=zoomActions, saveLabel=saveLabel,
undo=undo, undoLastPoint=undoLastPoint,
fileMenuActions=(
opendir, saveLabel, resetAll, quit),
beginner=(), advanced=(),
editMenu=(createpoly, edit, copy, delete,singleRere,
editMenu=(createpoly, edit, copy, delete,singleRere,None, undo, undoLastPoint,
None, color1, self.drawSquaresOption),
beginnerContext=(create, edit, copy, delete, singleRere),
advancedContext=(createMode, editMode, edit, copy,
......@@ -549,8 +556,13 @@ class MainWindow(QMainWindow, WindowMixin):
self.labelDialogOption.setChecked(settings.get(SETTING_PAINT_LABEL, False))
self.labelDialogOption.triggered.connect(self.speedChoose)
self.autoSaveOption = QAction(getStr('autoSaveMode'), self)
self.autoSaveOption.setCheckable(True)
self.autoSaveOption.setChecked(settings.get(SETTING_PAINT_LABEL, False))
self.autoSaveOption.triggered.connect(self.autoSaveFunc)
addActions(self.menus.file,
(opendir, None, saveLabel, saveRec, None, resetAll, deleteImg, quit))
(opendir, None, saveLabel, saveRec, self.autoSaveOption, None, resetAll, deleteImg, quit))
addActions(self.menus.help, (showSteps, showInfo))
addActions(self.menus.view, (
......@@ -566,9 +578,9 @@ class MainWindow(QMainWindow, WindowMixin):
# Custom context menu for the canvas widget:
addActions(self.canvas.menus[0], self.actions.beginnerContext)
addActions(self.canvas.menus[1], (
action('&Copy here', self.copyShape),
action('&Move here', self.moveShape)))
#addActions(self.canvas.menus[1], (
# action('&Copy here', self.copyShape),
# action('&Move here', self.moveShape)))
self.statusBar().showMessage('%s started.' % __appname__)
......@@ -758,6 +770,7 @@ class MainWindow(QMainWindow, WindowMixin):
self.canvas.setEditing(False)
self.canvas.fourpoint = True
self.actions.create.setEnabled(False)
self.actions.undoLastPoint.setEnabled(True)
def toggleDrawingSensitive(self, drawing=True):
"""In the middle of drawing, toggling between modes should be disabled."""
......@@ -866,10 +879,11 @@ class MainWindow(QMainWindow, WindowMixin):
self.updateComboBox()
def updateBoxlist(self):
shape = self.canvas.selectedShape
for shape in self.canvas.selectedShapes+[self.canvas.hShape]:
item = self.shapesToItemsbox[shape] # listitem
text = [(int(p.x()), int(p.y())) for p in shape.points]
item.setText(str(text))
self.actions.undo.setEnabled(True)
self.setDirty()
def indexTo5Files(self, currIndex):
......@@ -902,23 +916,27 @@ class MainWindow(QMainWindow, WindowMixin):
if len(self.mImgList) > 0:
self.zoomWidget.setValue(self.zoomWidgetValue + self.imgsplider.value())
# React to canvas signals.
def shapeSelectionChanged(self, selected=False):
if self._noSelectionSlot:
self._noSelectionSlot = False
else:
shape = self.canvas.selectedShape
if shape:
self.shapesToItems[shape].setSelected(True)
self.shapesToItemsbox[shape].setSelected(True) # ADD
else:
def shapeSelectionChanged(self, selected_shapes):
self._noSelectionSlot = True
for shape in self.canvas.selectedShapes:
shape.selected = False
self.labelList.clearSelection()
self.actions.delete.setEnabled(selected)
self.actions.copy.setEnabled(selected)
self.actions.edit.setEnabled(selected)
self.actions.shapeLineColor.setEnabled(selected)
self.actions.shapeFillColor.setEnabled(selected)
self.actions.singleRere.setEnabled(selected)
self.canvas.selectedShapes = selected_shapes
for shape in self.canvas.selectedShapes:
shape.selected = True
self.shapesToItems[shape].setSelected(True)
self.shapesToItemsbox[shape].setSelected(True)
self.labelList.scrollToItem(self.currentItem()) # QAbstractItemView.EnsureVisible
self.BoxList.scrollToItem(self.currentBox())
self._noSelectionSlot = False
n_selected = len(selected_shapes)
self.actions.singleRere.setEnabled(n_selected)
self.actions.delete.setEnabled(n_selected)
self.actions.copy.setEnabled(n_selected)
self.actions.edit.setEnabled(n_selected == 1)
def addLabel(self, shape):
shape.paintLabel = self.displayLabelOption.isChecked()
......@@ -941,10 +959,11 @@ class MainWindow(QMainWindow, WindowMixin):
action.setEnabled(True)
self.updateComboBox()
def remLabel(self, shape):
if shape is None:
def remLabels(self, shapes):
if shapes is None:
# print('rm empty label')
return
for shape in shapes:
item = self.shapesToItems[shape]
self.labelList.takeItem(self.labelList.row(item))
del self.shapesToItems[shape]
......@@ -1001,7 +1020,7 @@ class MainWindow(QMainWindow, WindowMixin):
item.setText(str([(int(p.x()), int(p.y())) for p in shape.points]))
self.updateComboBox()
def updateComboBox(self):
def updateComboBox(self): # TODO：貌似没用
# Get the unique labels and add them to the Combobox.
itemsTextList = [str(self.labelList.item(i).text()) for i in range(self.labelList.count())]
......@@ -1054,26 +1073,38 @@ class MainWindow(QMainWindow, WindowMixin):
return False
def copySelectedShape(self):
self.addLabel(self.canvas.copySelectedShape())
for shape in self.canvas.copySelectedShape():
self.addLabel(shape)
# fix copy and delete
self.shapeSelectionChanged(True)
#self.shapeSelectionChanged(True)
def labelSelectionChanged(self):
item = self.currentItem()
self.labelList.scrollToItem(item, QAbstractItemView.EnsureVisible)
if item and self.canvas.editing():
self._noSelectionSlot = True
self.canvas.selectShape(self.itemsToShapes[item])
shape = self.itemsToShapes[item]
if self._noSelectionSlot:
return
if self.canvas.editing():
selected_shapes = []
for item in self.labelList.selectedItems():
selected_shapes.append(self.itemsToShapes[item])
if selected_shapes:
self.canvas.selectShapes(selected_shapes)
else:
self.canvas.deSelectShape()
def boxSelectionChanged(self):
item = self.currentBox()
self.BoxList.scrollToItem(item, QAbstractItemView.EnsureVisible)
if item and self.canvas.editing():
self._noSelectionSlot = True
self.canvas.selectShape(self.itemsToShapesbox[item])
shape = self.itemsToShapesbox[item]
if self._noSelectionSlot:
#self.BoxList.scrollToItem(self.currentBox(), QAbstractItemView.PositionAtCenter)
return
if self.canvas.editing():
selected_shapes = []
for item in self.BoxList.selectedItems():
selected_shapes.append(self.itemsToShapesbox[item])
if selected_shapes:
self.canvas.selectShapes(selected_shapes)
else:
self.canvas.deSelectShape()
def labelItemChanged(self, item):
shape = self.itemsToShapes[item]
......@@ -1113,6 +1144,8 @@ class MainWindow(QMainWindow, WindowMixin):
if self.beginner(): # Switch to edit mode.
self.canvas.setEditing(True)
self.actions.create.setEnabled(True)
self.actions.undoLastPoint.setEnabled(False)
self.actions.undo.setEnabled(True)
else:
self.actions.editMode.setEnabled(True)
self.setDirty()
......@@ -1548,6 +1581,7 @@ class MainWindow(QMainWindow, WindowMixin):
self.fileListWidget.insertItem(int(currIndex), item)
self.openNextImg()
self.actions.saveRec.setEnabled(True)
self.actions.saveLabel.setEnabled(True)
elif mode == 'Auto':
if annotationFilePath and self.saveLabels(annotationFilePath, mode=mode):
......@@ -1643,7 +1677,8 @@ class MainWindow(QMainWindow, WindowMixin):
self.setDirty()
def deleteSelectedShape(self):
self.remLabel(self.canvas.deleteSelected())
self.remLabels(self.canvas.deleteSelected())
self.actions.undo.setEnabled(True)
self.setDirty()
if self.noShapes():
for action in self.actions.onShapesPresent:
......@@ -1653,7 +1688,7 @@ class MainWindow(QMainWindow, WindowMixin):
color = self.colorDialog.getColor(self.lineColor, u'Choose line color',
default=DEFAULT_LINE_COLOR)
if color:
self.canvas.selectedShape.line_color = color
for shape in self.canvas.selectedShapes: shape.line_color = color
self.canvas.update()
self.setDirty()
......@@ -1661,7 +1696,7 @@ class MainWindow(QMainWindow, WindowMixin):
color = self.colorDialog.getColor(self.fillColor, u'Choose fill color',
default=DEFAULT_FILL_COLOR)
if color:
self.canvas.selectedShape.fill_color = color
for shape in self.canvas.selectedShapes: shape.fill_color = color
self.canvas.update()
self.setDirty()
......@@ -1785,7 +1820,7 @@ class MainWindow(QMainWindow, WindowMixin):
def singleRerecognition(self):
img = cv2.imread(self.filePath)
shape = self.canvas.selectedShape
for shape in self.canvas.selectedShapes:
box = [[int(p.x()), int(p.y())] for p in shape.points]
assert len(box) == 4
img_crop = get_rotate_crop_image(img, np.array(box, np.float32))
......@@ -1914,8 +1949,8 @@ class MainWindow(QMainWindow, WindowMixin):
self.savePPlabel()
def saveRecResult(self):
if None in [self.PPlabelpath, self.PPlabel, self.fileStatedict]:
QMessageBox.information(self, "Information", "Save file first")
if {} in [self.PPlabelpath, self.PPlabel, self.fileStatedict]:
QMessageBox.information(self, "Information", "Check the image first")
return
rec_gt_dir = os.path.dirname(self.PPlabelpath) + '/rec_gt.txt'
......@@ -1953,6 +1988,33 @@ class MainWindow(QMainWindow, WindowMixin):
self.canvas.newShape.disconnect()
self.canvas.newShape.connect(partial(self.newShape, False))
def autoSaveFunc(self):
if self.autoSaveOption.isChecked():
self.autoSaveNum = 1 # Real auto_Save
try:
self.saveLabelFile()
except:
pass
print('The program will automatically save once after confirming an image')
else:
self.autoSaveNum = 5 # Used for backup
print('The program will automatically save once after confirming 5 images (default)')
def undoShapeEdit(self):
self.canvas.restoreShape()
self.labelList.clear()
self.BoxList.clear()
self.loadShapes(self.canvas.shapes)
self.actions.undo.setEnabled(self.canvas.isShapeRestorable)
def loadShapes(self, shapes, replace=True):
self._noSelectionSlot = True
for shape in shapes:
self.addLabel(shape)
self.labelList.clearSelection()
self._noSelectionSlot = False
self.canvas.loadShapes(shapes, replace=replace)
def inverted(color):
return QColor(*[255 - v for v in color.getRgb()])
......
PPOCRLabel/README.md
浏览文件 @ 6ec9aa4d
......@@ -8,6 +8,10 @@ PPOCRLabel is a semi-automatic graphic annotation tool suitable for OCR field, w
### Recent Update
- 2021.2.5: New batch processing and undo functions (by [Evezerest](https://github.com/Evezerest)):
- Batch processing function: Press and hold the Ctrl key to select the box, you can move, copy, and delete in batches.
- Undo function: In the process of drawing a four-point label box or after editing the box, press Ctrl+Z to undo the previous operation.
- Fix image rotation and size problems, optimize the process of editing the mark frame (by [ninetailskim](https://github.com/ninetailskim)、 [edencfc](https://github.com/edencfc)).
- 2021.1.11: Optimize the labeling experience (by [edencfc](https://github.com/edencfc)),
- Users can choose whether to pop up the label input dialog after drawing the detection box in "View - Pop-up Label Input Dialog".
- The recognition result scrolls synchronously when users click related detection box.
......@@ -16,7 +20,6 @@ PPOCRLabel is a semi-automatic graphic annotation tool suitable for OCR field, w
### TODO:
- Lock box mode: For the same scene data, the size and position of the locked detection box can be transferred between different pictures.
- Experience optimization: Add undo, batch operation include move, copy, delete and so on, optimize the annotation process.
## Installation
......@@ -49,7 +52,7 @@ python3 PPOCRLabel.py
```
pip3 install pyqt5
pip3 uninstall opencv-python # Uninstall opencv manually as it conflicts with pyqt
pip3 install opencv-contrib-python-headless # Install the headless version of opencv
pip3 install opencv-contrib-python-headless==4.2.0.32 # Install the headless version of opencv
cd ./PPOCRLabel # Change the directory to the PPOCRLabel folder
python3 PPOCRLabel.py
```
......@@ -76,12 +79,11 @@ python3 PPOCRLabel.py
7. Double click the result in 'recognition result' list to manually change inaccurate recognition results.
8. Click "Check", the image status will switch to "√",then the program automatically jump to the next(The results will not be written directly to the file at this time).
8. Click "Check", the image status will switch to "√",then the program automatically jump to the next.
9. Click "Delete Image" and the image will be deleted to the recycle bin.
10. Labeling result: the user can save manually through the menu "File - Save Label", while the program will also save automatically after every 5 images confirmed by the user.the manually checked label will be stored in *Label.txt* under the opened picture folder.
Click "PaddleOCR"-"Save Recognition Results" in the menu bar, the recognition training data of such pictures will be saved in the *crop_img* folder, and the recognition label will be saved in *rec_gt.txt*<sup>[4]</sup>.
10. Labeling result: the user can save manually through the menu "File - Save Label", while the program will also save automatically if "File - Auto Save Label Mode" is selected. The manually checked label will be stored in *Label.txt* under the opened picture folder. Click "PaddleOCR"-"Save Recognition Results" in the menu bar, the recognition training data of such pictures will be saved in the *crop_img* folder, and the recognition label will be saved in *rec_gt.txt*<sup>[4]</sup>.
### Note
......@@ -89,8 +91,7 @@ python3 PPOCRLabel.py
[2] The image status indicates whether the user has saved the image manually. If it has not been saved manually it is "X", otherwise it is "√", PPOCRLabel will not relabel pictures with a status of "√".
[3] After clicking "Re-recognize", the model will overwrite ALL recognition results in the picture.
Therefore, if the recognition result has been manually changed before, it may change after re-recognition.
[3] After clicking "Re-recognize", the model will overwrite ALL recognition results in the picture. Therefore, if the recognition result has been manually changed before, it may change after re-recognition.
[4] The files produced by PPOCRLabel can be found under the opened picture folder including the following, please do not manually change the contents, otherwise it will cause the program to be abnormal.
......@@ -107,13 +108,14 @@ Therefore, if the recognition result has been manually changed before, it may ch
### Shortcut keys
| Shortcut keys | Description |
| ---------------- | ------------------------------------------------ |
| Ctrl + shift + A | Automatically label all unchecked images |
| Ctrl + shift + R | Re-recognize all the labels of the current image |
| ------------------------ | ------------------------------------------------ |
| Ctrl + Shift + R | Re-recognize all the labels of the current image |
| W | Create a rect box |
| Q | Create a four-points box |
| Ctrl + E | Edit label of the selected box |
| Ctrl + R | Re-recognize the selected box |
| Ctrl + C | Copy and paste the selected box |
| Ctrl + Left Mouse Button | Multi select the label box |
| Backspace | Delete the selected box |
| Ctrl + V | Check image |
| Ctrl + Shift + d | Delete image |
......@@ -136,7 +138,7 @@ Therefore, if the recognition result has been manually changed before, it may ch
PPOCRLabel supports three ways to save Label.txt
- Automatically save: When it detects that the user has manually checked 5 pictures, the program automatically writes the annotations into Label.txt. The user can change the value of ``self.autoSaveNum`` in ``PPOCRLabel.py`` to set the number of images to be automatically saved after confirmation.
- Automatically save: After selecting "File - Auto Save Label Mode", the program will automatically write the annotations into Label.txt every time the user confirms an image. If this option is not turned on, it will be automatically saved after detecting that the user has manually checked 5 images.
- Manual save: Click "File-Save Marking Results" to manually save the label.
- Close application save
......@@ -160,9 +162,9 @@ For some data that are difficult to recognize, the recognition results will not
```
pyrcc5 -o libs/resources.py resources.qrc
```
- If you get an error ``` module 'cv2' has no attribute 'INTER_NEAREST'```, you need to delete all opencv related packages first, and then reinstall the headless version of opencv
- If you get an error ``` module 'cv2' has no attribute 'INTER_NEAREST'```, you need to delete all opencv related packages first, and then reinstall the 4.2.0.32 version of headless opencv
```
pip install opencv-contrib-python-headless
pip install opencv-contrib-python-headless==4.2.0.32
```
### Related
......
PPOCRLabel/README_ch.md
浏览文件 @ 6ec9aa4d
......@@ -8,6 +8,10 @@ PPOCRLabel是一款适用于OCR领域的半自动化图形标注工具，内置P
#### 近期更新
- 2021.2.5：新增批处理与撤销功能（by [Evezerest](https://github.com/Evezerest))
- 批处理功能：按住Ctrl键选择标记框后可批量移动、复制、删除。
- 撤销功能：在绘制四点标注框过程中或对框进行编辑操作后，按下Ctrl+Z可撤销上一部操作。
- 修复图像旋转和尺寸问题、优化编辑标记框过程（by [ninetailskim](https://github.com/ninetailskim)、 [edencfc](https://github.com/edencfc)）
- 2021.1.11：优化标注体验（by [edencfc](https://github.com/edencfc)）：
- 用户可在“视图 - 弹出标记输入框”选择在画完检测框后标记输入框是否弹出。
- 识别结果与检测框同步滚动。
......@@ -17,9 +21,8 @@ PPOCRLabel是一款适用于OCR领域的半自动化图形标注工具，内置P
#### 尽请期待
- 锁定框模式：针对同一场景数据，被锁定的检测框的大小与位置能在不同图片之间传递。
- 体验优化：增加撤销操作，批量移动、复制、删除等功能。优化标注流程。
如果您对以上内容感兴趣或对完善工具有不一样的想法，欢迎加入我们的队伍与我们共同开发
如果您对以上内容感兴趣或对完善工具有不一样的想法，欢迎加入我们的SIG队伍与我们共同开发。可以在[此处](https://github.com/PaddlePaddle/PaddleOCR/issues/1728)完成问卷和前置任务，经过我们确认相关内容后即可正式加入，享受SIG福利，共同为OCR开源事业贡献（特别说明：针对PPOCRLabel的改进也属于PaddleOCR前置任务）
## 安装
......@@ -49,7 +52,7 @@ python3 PPOCRLabel.py --lang ch
```
pip3 install pyqt5
pip3 uninstall opencv-python # 由于mac版本的opencv与pyqt有冲突，需先手动卸载opencv
pip3 install opencv-contrib-python-headless # 安装headless版本的open-cv
pip3 install opencv-contrib-python-headless==4.2.0.32 # 安装headless版本的open-cv
cd ./PPOCRLabel # 将目录切换到PPOCRLabel文件夹下
python3 PPOCRLabel.py --lang ch
```
......@@ -65,9 +68,9 @@ python3 PPOCRLabel.py --lang ch
5. 标记框绘制完成后，用户点击 “确认”，检测框会先被预分配一个 “待识别” 标签。
6. 重新识别：将图片中的所有检测画绘制/调整完成后，点击 “重新识别”，PPOCR模型会对当前图片中的**所有检测框**重新识别<sup>[3]</sup>。
7. 内容更改：双击识别结果，对不准确的识别结果进行手动更改。
8. 确认标记：点击 “确认”，图片状态切换为 “√”，跳转至下一张（此时不会直接将结果写入文件）。
8. **确认标记**：点击 “确认”，图片状态切换为 “√”，跳转至下一张。
9. 删除：点击 “删除图像”，图片将会被删除至回收站。
10. 保存结果：用户可以通过菜单中“文件-保存标记结果”手动保存，同时程序也会在用户每确认5张图片后自动保存一次。手动确认过的标记将会被存放在所打开图片文件夹下的*Label.txt*中。在菜单栏点击 “文件” - "保存识别结果"后，会将此类图片的识别训练数据保存在*crop_img*文件夹下，识别标签保存在*rec_gt.txt*中<sup>[4]</sup>。
10. 保存结果：用户可以通过菜单中“文件-保存标记结果”手动保存，同时也可以点击“文件 - 自动保存标记结果”开启自动保存。手动确认过的标记将会被存放在所打开图片文件夹下的*Label.txt*中。在菜单栏点击 “文件” - "保存识别结果"后，会将此类图片的识别训练数据保存在*crop_img*文件夹下，识别标签保存在*rec_gt.txt*中<sup>[4]</sup>。
### 注意
......@@ -93,12 +96,13 @@ python3 PPOCRLabel.py --lang ch
| 快捷键 | 说明 |
| ---------------- | ---------------------------- |
| Ctrl + shift + A | 自动标注所有未确认过的图片 |
| Ctrl + shift + R | 对当前图片的所有标记重新识别 |
| W | 新建矩形框 |
| Q | 新建四点框 |
| Ctrl + E | 编辑所选框标签 |
| Ctrl + R | 重新识别所选标记 |
| Ctrl + C | 复制并粘贴选中的标记框 |
| Ctrl + 鼠标左键 | 多选标记框 |
| Backspace | 删除所选框 |
| Ctrl + V | 确认本张图片标记 |
| Ctrl + Shift + d | 删除本张图片 |
......@@ -120,7 +124,7 @@ python3 PPOCRLabel.py --lang ch
PPOCRLabel支持三种保存方式：
- 程序自动保存：当检测到用户手动确认过5张图片后，程序自动将标记结果写入Label.txt中。其中用户可通过更改```PPOCRLabel.py```中的```self.autoSaveNum```的数值设置确认几张图片后进行自动保存。
- 自动保存：点击“文件 - 自动保存标记结果”后，用户每确认过一张图片，程序自动将标记结果写入Label.txt中。若未开启此选项，则检测到用户手动确认过5张图片后进行自动保存。
- 手动保存：点击“文件 - 保存标记结果”手动保存标记。
- 关闭应用程序保存
......@@ -145,9 +149,9 @@ PPOCRLabel支持三种保存方式：
pyrcc5 -o libs/resources.py resources.qrc
```
- 如果出现``` module 'cv2' has no attribute 'INTER_NEAREST'```错误，需要首先删除所有opencv相关包，然后重新安装headless版本的opencv
- 如果出现``` module 'cv2' has no attribute 'INTER_NEAREST'```错误，需要首先删除所有opencv相关包，然后重新安装4.2.0.32版本的headless opencv
```
pip install opencv-contrib-python-headless
pip install opencv-contrib-python-headless==4.2.0.32
```
### 参考资料
......
PPOCRLabel/libs/canvas.py
浏览文件 @ 6ec9aa4d
......@@ -37,7 +37,8 @@ class Canvas(QWidget):
zoomRequest = pyqtSignal(int)
scrollRequest = pyqtSignal(int, int)
newShape = pyqtSignal()
selectionChanged = pyqtSignal(bool)
# selectionChanged = pyqtSignal(bool)
selectionChanged = pyqtSignal(list)
shapeMoved = pyqtSignal()
drawingPolygon = pyqtSignal(bool)
......@@ -51,9 +52,11 @@ class Canvas(QWidget):
# Initialise local state.
self.mode = self.EDIT
self.shapes = []
self.shapesBackups = []
self.current = None
self.selectedShapes = []
self.selectedShape = None # save the selected shape here
self.selectedShapeCopy = None
self.selectedShapesCopy = []
self.drawingLineColor = QColor(0, 0, 255)
self.drawingRectColor = QColor(0, 0, 255)
self.line = Shape(line_color=self.drawingLineColor)
......@@ -77,6 +80,7 @@ class Canvas(QWidget):
self.drawSquare = False
self.fourpoint = True # ADD
self.pointnum = 0
self.movingShape = False
#initialisation for panning
self.pan_initial_pos = QPoint()
......@@ -149,37 +153,20 @@ class Canvas(QWidget):
clipped_x = min(max(0, pos.x()), size.width())
clipped_y = min(max(0, pos.y()), size.height())
pos = QPointF(clipped_x, clipped_y)
elif len(self.current) > 1 and self.closeEnough(pos, self.current[0]) and not self.fourpoint:
elif len(self.current) > 1 and self.closeEnough(pos, self.current[0]):
# Attract line to starting point and colorise to alert the
# user:
pos = self.current[0]
color = self.current.line_color
self.overrideCursor(CURSOR_POINT)
self.current.highlightVertex(0, Shape.NEAR_VERTEX)
elif ( # ADD
len(self.current) > 1
and self.fourpoint
and self.closeEnough(pos, self.current[0])
):
# Attract line to starting point and
# colorise to alert the user.
pos = self.current[0]
self.overrideCursor(CURSOR_POINT)
self.current.highlightVertex(0, Shape.NEAR_VERTEX)
if self.drawSquare:
initPos = self.current[0]
minX = initPos.x()
minY = initPos.y()
min_size = min(abs(pos.x() - minX), abs(pos.y() - minY))
directionX = -1 if pos.x() - minX < 0 else 1
directionY = -1 if pos.y() - minY < 0 else 1
self.line[1] = QPointF(minX + directionX * min_size, minY + directionY * min_size)
self.line.points = [self.current[0], pos]
self.line.close()
elif self.fourpoint:
# self.line[self.pointnum] = pos # OLD
self.line[0] = self.current[-1]
self.line[1] = pos
......@@ -196,12 +183,14 @@ class Canvas(QWidget):
# Polygon copy moving.
if Qt.RightButton & ev.buttons():
if self.selectedShapeCopy and self.prevPoint:
if self.selectedShapesCopy and self.prevPoint:
self.overrideCursor(CURSOR_MOVE)
self.boundedMoveShape(self.selectedShapeCopy, pos)
self.boundedMoveShape(self.selectedShapesCopy, pos)
self.repaint()
elif self.selectedShape:
self.selectedShapeCopy = self.selectedShape.copy()
elif self.selectedShapes:
self.selectedShapesCopy = [
s.copy() for s in self.selectedShapes
]
self.repaint()
return
......@@ -211,11 +200,13 @@ class Canvas(QWidget):
self.boundedMoveVertex(pos)
self.shapeMoved.emit()
self.repaint()
elif self.selectedShape and self.prevPoint:
self.movingShape = True
elif self.selectedShapes and self.prevPoint:
self.overrideCursor(CURSOR_MOVE)
self.boundedMoveShape(self.selectedShape, pos)
self.boundedMoveShape(self.selectedShapes, pos)
self.shapeMoved.emit()
self.repaint()
self.movingShape = True
else:
#pan
delta_x = pos.x() - self.pan_initial_pos.x()
......@@ -263,65 +254,60 @@ class Canvas(QWidget):
def mousePressEvent(self, ev):
pos = self.transformPos(ev.pos())
if ev.button() == Qt.LeftButton:
if self.drawing():
# self.handleDrawing(pos) # OLD
if self.current and self.fourpoint: # ADD IF
if self.current:
if self.fourpoint: # ADD IF
# Add point to existing shape.
print('Adding points in mousePressEvent is ', self.line[1])
# print('Adding points in mousePressEvent is ', self.line[1])
self.current.addPoint(self.line[1])
self.line[0] = self.current[-1]
if self.current.isClosed():
# print('1111')
self.finalise()
elif self.drawSquare: # 增加
assert len(self.current.points) == 1
self.current.points = self.line.points
self.finalise()
elif not self.outOfPixmap(pos):
# Create new shape.
self.current = Shape()# self.current = Shape(shape_type=self.createMode)
self.current = Shape()
self.current.addPoint(pos)
# if self.createMode == "point":
# self.finalise()
# else:
# if self.createMode == "circle":
# self.current.shape_type = "circle"
self.line.points = [pos, pos]
self.setHiding()
self.drawingPolygon.emit(True)
self.update()
else:
selection = self.selectShapePoint(pos)
group_mode = int(ev.modifiers()) == Qt.ControlModifier
self.selectShapePoint(pos, multiple_selection_mode=group_mode)
self.prevPoint = pos
if selection is None:
#pan
QApplication.setOverrideCursor(QCursor(Qt.OpenHandCursor))
self.pan_initial_pos = pos
elif ev.button() == Qt.RightButton and self.editing():
self.selectShapePoint(pos)
group_mode = int(ev.modifiers()) == Qt.ControlModifier
self.selectShapePoint(pos, multiple_selection_mode=group_mode)
self.prevPoint = pos
self.update()
def mouseReleaseEvent(self, ev):
if ev.button() == Qt.RightButton:
menu = self.menus[bool(self.selectedShapeCopy)]
menu = self.menus[bool(self.selectedShapesCopy)]
self.restoreCursor()
if not menu.exec_(self.mapToGlobal(ev.pos()))\
and self.selectedShapeCopy:
and self.selectedShapesCopy:
# Cancel the move by deleting the shadow copy.
self.selectedShapeCopy = None
# self.selectedShapeCopy = None
self.selectedShapesCopy = []
self.repaint()
elif ev.button() == Qt.LeftButton and self.selectedShape: # OLD
elif ev.button() == Qt.LeftButton and self.selectedShapes:
if self.selectedVertex():
self.overrideCursor(CURSOR_POINT)
else:
self.overrideCursor(CURSOR_GRAB)
elif ev.button() == Qt.LeftButton and not self.fourpoint:
pos = self.transformPos(ev.pos())
if self.drawing():
......@@ -330,24 +316,37 @@ class Canvas(QWidget):
#pan
QApplication.restoreOverrideCursor() # ?
if self.movingShape and self.hShape:
index = self.shapes.index(self.hShape)
if (
self.shapesBackups[-1][index].points
!= self.shapes[index].points
):
self.storeShapes()
self.shapeMoved.emit() # connect to updateBoxlist in PPOCRLabel.py
self.movingShape = False
def endMove(self, copy=False):
assert self.selectedShape and self.selectedShapeCopy
shape = self.selectedShapeCopy
#del shape.fill_color
#del shape.line_color
assert self.selectedShapes and self.selectedShapesCopy
assert len(self.selectedShapesCopy) == len(self.selectedShapes)
if copy:
for i, shape in enumerate(self.selectedShapesCopy):
self.shapes.append(shape)
self.selectedShape.selected = False
self.selectedShape = shape
self.repaint()
self.selectedShapes[i].selected = False
self.selectedShapes[i] = shape
else:
self.selectedShape.points = [p for p in shape.points]
self.selectedShapeCopy = None
for i, shape in enumerate(self.selectedShapesCopy):
self.selectedShapes[i].points = shape.points
self.selectedShapesCopy = []
self.repaint()
self.storeShapes()
return True
def hideBackroundShapes(self, value):
self.hideBackround = value
if self.selectedShape:
if self.selectedShapes:
# Only hide other shapes if there is a current selection.
# Otherwise the user will not be able to select a shape.
self.setHiding(True)
......@@ -363,7 +362,7 @@ class Canvas(QWidget):
if self.pointnum == 3:
self.finalise()
else: # 按住送掉后跳到这里
else:
initPos = self.current[0]
print('initPos', self.current[0])
minX = initPos.x()
......@@ -399,28 +398,33 @@ class Canvas(QWidget):
self.current.popPoint()
self.finalise()
def selectShape(self, shape):
self.deSelectShape()
shape.selected = True
self.selectedShape = shape
def selectShapes(self, shapes):
for s in shapes: s.seleted = True
self.setHiding()
self.selectionChanged.emit(True)
self.selectionChanged.emit(shapes)
self.update()
def selectShapePoint(self, point):
def selectShapePoint(self, point, multiple_selection_mode):
"""Select the first shape created which contains this point."""
self.deSelectShape()
if self.selectedVertex(): # A vertex is marked for selection.
index, shape = self.hVertex, self.hShape
shape.highlightVertex(index, shape.MOVE_VERTEX)
self.selectShape(shape)
return self.hVertex
else:
for shape in reversed(self.shapes):
if self.isVisible(shape) and shape.containsPoint(point):
self.selectShape(shape)
self.calculateOffsets(shape, point)
return self.selectedShape
return None
self.setHiding()
if multiple_selection_mode:
if shape not in self.selectedShapes: # list
self.selectionChanged.emit(
self.selectedShapes + [shape]
)
else:
self.selectionChanged.emit([shape])
return
self.deSelectShape()
def calculateOffsets(self, shape, point):
rect = shape.boundingRect()
......@@ -465,8 +469,9 @@ class Canvas(QWidget):
else:
shiftPos = pos - point
if [shape[0].x(), shape[0].y(), shape[2].x(), shape[2].y()] \
== [shape[3].x(),shape[1].y(),shape[1].x(),shape[3].y()]:
shape.moveVertexBy(index, shiftPos)
lindex = (index + 1) % 4
rindex = (index + 3) % 4
lshift = None
......@@ -480,7 +485,12 @@ class Canvas(QWidget):
shape.moveVertexBy(rindex, rshift)
shape.moveVertexBy(lindex, lshift)
def boundedMoveShape(self, shape, pos):
else:
shape.moveVertexBy(index, shiftPos)
def boundedMoveShape(self, shapes, pos):
if type(shapes).__name__ != 'list': shapes = [shapes]
if self.outOfPixmap(pos):
return False # No need to move
o1 = pos + self.offsets[0]
......@@ -497,40 +507,49 @@ class Canvas(QWidget):
#self.calculateOffsets(self.selectedShape, pos)
dp = pos - self.prevPoint
if dp:
for shape in shapes:
shape.moveBy(dp)
self.prevPoint = pos
return True
return False
def deSelectShape(self):
if self.selectedShape:
self.selectedShape.selected = False
self.selectedShape = None
if self.selectedShapes:
for shape in self.selectedShapes: shape.selected=False
self.setHiding(False)
self.selectionChanged.emit(False)
self.selectionChanged.emit([])
self.update()
def deleteSelected(self):
if self.selectedShape:
shape = self.selectedShape
self.shapes.remove(self.selectedShape)
self.selectedShape = None
deleted_shapes = []
if self.selectedShapes:
for shape in self.selectedShapes:
self.shapes.remove(shape)
deleted_shapes.append(shape)
self.storeShapes()
self.selectedShapes = []
self.update()
return shape
return deleted_shapes
def storeShapes(self):
shapesBackup = []
for shape in self.shapes:
shapesBackup.append(shape.copy())
if len(self.shapesBackups) >= 10:
self.shapesBackups = self.shapesBackups[-9:]
self.shapesBackups.append(shapesBackup)
def copySelectedShape(self):
if self.selectedShape:
shape = self.selectedShape.copy()
self.deSelectShape()
self.shapes.append(shape)
shape.selected = True
self.selectedShape = shape
self.boundedShiftShape(shape)
return shape
if self.selectedShapes:
self.selectedShapesCopy = [s.copy() for s in self.selectedShapes]
self.boundedShiftShapes(self.selectedShapesCopy)
self.endMove(copy=True)
return self.selectedShapes
def boundedShiftShape(self, shape):
def boundedShiftShapes(self, shapes):
# Try to move in one direction, and if it fails in another.
# Give up if both fail.
for shape in shapes:
point = shape[0]
offset = QPointF(2.0, 2.0)
self.calculateOffsets(shape, point)
......@@ -560,8 +579,9 @@ class Canvas(QWidget):
if self.current:
self.current.paint(p)
self.line.paint(p)
if self.selectedShapeCopy:
self.selectedShapeCopy.paint(p)
if self.selectedShapesCopy:
for s in self.selectedShapesCopy:
s.paint(p)
# Paint rect
if self.current is not None and len(self.line) == 2 and not self.fourpoint:
......@@ -739,6 +759,7 @@ class Canvas(QWidget):
if fill_color:
self.shapes[-1].fill_color = fill_color
self.storeShapes()
return self.shapes[-1]
......@@ -749,6 +770,17 @@ class Canvas(QWidget):
self.line.points = [self.current[-1], self.current[0]]
self.drawingPolygon.emit(True)
def undoLastPoint(self):
if not self.current or self.current.isClosed():
return
self.current.popPoint()
if len(self.current) > 0:
self.line[0] = self.current[-1]
else:
self.current = None
self.drawingPolygon.emit(False)
self.repaint()
def resetAllLines(self):
assert self.shapes
self.current = self.shapes.pop()
......@@ -762,11 +794,18 @@ class Canvas(QWidget):
def loadPixmap(self, pixmap):
self.pixmap = pixmap
self.shapes = []
self.repaint() # 这函数在哪
self.repaint()
def loadShapes(self, shapes):
def loadShapes(self, shapes, replace=True):
if replace:
self.shapes = list(shapes)
else:
self.shapes.extend(shapes)
self.current = None
self.hShape = None
self.hVertex = None
# self.hEdge = None
self.storeShapes()
self.repaint()
def setShapeVisible(self, shape, value):
......@@ -793,6 +832,24 @@ class Canvas(QWidget):
self.restoreCursor()
self.pixmap = None
self.update()
self.shapesBackups = []
def setDrawingShapeToSquare(self, status):
self.drawSquare = status
def restoreShape(self):
if not self.isShapeRestorable:
return
self.shapesBackups.pop() # latest
shapesBackup = self.shapesBackups.pop()
self.shapes = shapesBackup
self.selectedShapes = []
for shape in self.shapes:
shape.selected = False
self.repaint()
@property
def isShapeRestorable(self):
if len(self.shapesBackups) < 2:
return False
return True
\ No newline at end of file
PPOCRLabel/libs/resources.py
浏览文件 @ 6ec9aa4d
因为 它太大了无法显示 source diff 。你可以改为 查看blob。
PPOCRLabel/libs/shape.py
浏览文件 @ 6ec9aa4d
......@@ -82,7 +82,7 @@ class Shape(object):
return False
def addPoint(self, point):
if not self.reachMaxPoints():
if not self.reachMaxPoints(): # 4个点时发出close信号
self.points.append(point)
def popPoint(self):
......
PPOCRLabel/resources/strings/strings-zh-CN.properties
浏览文件 @ 6ec9aa4d
......@@ -97,3 +97,6 @@ showBox=显示所有标注
saveLabel=保存标记结果
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=自动保存标记结果
\ No newline at end of file
PPOCRLabel/resources/strings/strings.properties
浏览文件 @ 6ec9aa4d
......@@ -97,3 +97,6 @@ showBox=Show All Box
saveLabel=Save Label
singleRe=Re-recognition RectBox
labelDialogOption=Pop-up Label Input Dialog
undo=Undo
undoLastPoint=Undo Last Point
autoSaveMode=Auto Save Label Mode
\ No newline at end of file
configs/cls/cls_mv3.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [0, 1000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......
configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,10 @@ Global:
save_epoch_step: 1200
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [3000, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
# 1. If pretrained_model is saved in static mode, such as classification pretrained model
# from static branch, load_static_weights must be set as True.
# 2. If you want to finetune the pretrained models we provide in the docs,
# you should set load_static_weights as False.
load_static_weights: True
cal_metric_during_train: False
pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
......
configs/det/ch_ppocr_v2.0/ch_det_res18_db_v2.0.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,10 @@ Global:
save_epoch_step: 1200
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [3000, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
# 1. If pretrained_model is saved in static mode, such as classification pretrained model
# from static branch, load_static_weights must be set as True.
# 2. If you want to finetune the pretrained models we provide in the docs,
# you should set load_static_weights as False.
load_static_weights: True
cal_metric_during_train: False
pretrained_model: ./pretrain_models/ResNet18_vd_pretrained
......
configs/det/det_mv3_db.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,10 @@ Global:
save_epoch_step: 1200
# evaluation is run every 2000 iterations
eval_batch_step: [0, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
# 1. If pretrained_model is saved in static mode, such as classification pretrained model
# from static branch, load_static_weights must be set as True.
# 2. If you want to finetune the pretrained models we provide in the docs,
# you should set load_static_weights as False.
load_static_weights: True
cal_metric_during_train: False
pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
......
configs/det/det_mv3_east.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,10 @@ Global:
save_epoch_step: 1000
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [4000, 5000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
# 1. If pretrained_model is saved in static mode, such as classification pretrained model
# from static branch, load_static_weights must be set as True.
# 2. If you want to finetune the pretrained models we provide in the docs,
# you should set load_static_weights as False.
load_static_weights: True
cal_metric_during_train: False
pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
......
configs/det/det_r50_vd_db.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,10 @@ Global:
save_epoch_step: 1200
# evaluation is run every 2000 iterations
eval_batch_step: [0,2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
# 1. If pretrained_model is saved in static mode, such as classification pretrained model
# from static branch, load_static_weights must be set as True.
# 2. If you want to finetune the pretrained models we provide in the docs,
# you should set load_static_weights as False.
load_static_weights: True
cal_metric_during_train: False
pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained
......
configs/det/det_r50_vd_east.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,10 @@ Global:
save_epoch_step: 1000
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [4000, 5000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
# 1. If pretrained_model is saved in static mode, such as classification pretrained model
# from static branch, load_static_weights must be set as True.
# 2. If you want to finetune the pretrained models we provide in the docs,
# you should set load_static_weights as False.
load_static_weights: True
cal_metric_during_train: False
pretrained_model: ./pretrain_models/ResNet50_vd_pretrained/
......
configs/det/det_r50_vd_sast_icdar15.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,10 @@ Global:
save_epoch_step: 1000
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [4000, 5000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
# 1. If pretrained_model is saved in static mode, such as classification pretrained model
# from static branch, load_static_weights must be set as True.
# 2. If you want to finetune the pretrained models we provide in the docs,
# you should set load_static_weights as False.
load_static_weights: True
cal_metric_during_train: False
pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained/
......
configs/det/det_r50_vd_sast_totaltext.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,10 @@ Global:
save_epoch_step: 1000
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [4000, 5000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
# 1. If pretrained_model is saved in static mode, such as classification pretrained model
# from static branch, load_static_weights must be set as True.
# 2. If you want to finetune the pretrained models we provide in the docs,
# you should set load_static_weights as False.
load_static_weights: True
cal_metric_during_train: False
pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained/
......
configs/rec/ch_ppocr_v2.0/rec_chinese_common_train_v2.0.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [0, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......
configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [0, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......
configs/rec/rec_icdar15_train.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 2000 iterations
eval_batch_step: [0, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......
configs/rec/rec_mv3_none_bilstm_ctc.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 2000 iterations
eval_batch_step: [0, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......
configs/rec/rec_mv3_none_none_ctc.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 2000 iterations
eval_batch_step: [0, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......
configs/rec/rec_mv3_tps_bilstm_att.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [0, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......@@ -66,7 +65,7 @@ Metric:
Train:
dataset:
name: LMDBDataSet
data_dir: ../training/
data_dir: ./train_data/data_lmdb_release/training/
transforms:
- DecodeImage: # load image
img_mode: BGR
......@@ -85,7 +84,7 @@ Train:
Eval:
dataset:
name: LMDBDataSet
data_dir: ../validation/
data_dir: ./train_data/data_lmdb_release/validation/
transforms:
- DecodeImage: # load image
img_mode: BGR
......
configs/rec/rec_mv3_tps_bilstm_ctc.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 2000 iterations
eval_batch_step: [0, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......
configs/rec/rec_r34_vd_none_bilstm_ctc.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 2000 iterations
eval_batch_step: [0, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......
configs/rec/rec_r34_vd_none_none_ctc.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 2000 iterations
eval_batch_step: [0, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......
configs/rec/rec_r34_vd_tps_bilstm_att.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [0, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......@@ -65,7 +64,7 @@ Metric:
Train:
dataset:
name: LMDBDataSet
data_dir: ../training/
data_dir: ./train_data/data_lmdb_release/training/
transforms:
- DecodeImage: # load image
img_mode: BGR
......@@ -84,7 +83,7 @@ Train:
Eval:
dataset:
name: LMDBDataSet
data_dir: ../validation/
data_dir: ./train_data/data_lmdb_release/validation/
transforms:
- DecodeImage: # load image
img_mode: BGR
......
configs/rec/rec_r34_vd_tps_bilstm_ctc.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 2000 iterations
eval_batch_step: [0, 2000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......
configs/rec/rec_r50_fpn_srn.yml
浏览文件 @ 6ec9aa4d
......@@ -7,7 +7,6 @@ Global:
save_epoch_step: 3
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [0, 5000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......@@ -59,7 +58,7 @@ Metric:
Train:
dataset:
name: LMDBDataSet
data_dir: ./train_data/srn_train_data_duiqi
data_dir: ./train_data/data_lmdb_release/training/
transforms:
- DecodeImage: # load image
img_mode: BGR
......@@ -84,7 +83,7 @@ Train:
Eval:
dataset:
name: LMDBDataSet
data_dir: ./train_data/data_lmdb_release/evaluation
data_dir: ./train_data/data_lmdb_release/validation/
transforms:
- DecodeImage: # load image
img_mode: BGR
......
deploy/cpp_infer/CMakeLists.txt
浏览文件 @ 6ec9aa4d
......@@ -133,7 +133,11 @@ if(WITH_MKL)
endif ()
endif()
else()
if (WIN32)
set(MATH_LIB ${PADDLE_LIB}/third_party/install/openblas/lib/openblas${CMAKE_STATIC_LIBRARY_SUFFIX})
else ()
set(MATH_LIB ${PADDLE_LIB}/third_party/install/openblas/lib/libopenblas${CMAKE_STATIC_LIBRARY_SUFFIX})
endif ()
endif()
# Note: libpaddle_inference_api.so/a must put before libpaddle_fluid.so/a
......
deploy/cpp_infer/docs/windows_vs2019_build.md
浏览文件 @ 6ec9aa4d
......@@ -14,7 +14,7 @@ PaddleOCR在Windows 平台下基于`Visual Studio 2019 Community` 进行了测
### Step1: 下载PaddlePaddle C++ 预测库 fluid_inference
PaddlePaddle C++ 预测库针对不同的`CPU`和`CUDA`版本提供了不同的预编译版本，请根据实际情况下载: [C++预测库下载列表](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/advanced_guide/inference_deployment/inference/windows_cpp_inference.html)
PaddlePaddle C++ 预测库针对不同的`CPU`和`CUDA`版本提供了不同的预编译版本，请根据实际情况下载: [C++预测库下载列表](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/guides/05_inference_deployment/inference/windows_cpp_inference.html)
解压后`D:\projects\fluid_inference`目录包含内容为：
```
......
deploy/cpp_infer/readme.md
浏览文件 @ 6ec9aa4d
......@@ -74,9 +74,21 @@ opencv3/
* 有2种方式获取Paddle预测库，下面进行详细介绍。
#### 1.2.1 预测库源码编译
#### 1.2.1 直接下载安装
* [Paddle预测库官网](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/guides/05_inference_deployment/inference/build_and_install_lib_cn.html)上提供了不同cuda版本的Linux预测库，可以在官网查看并选择合适的预测库版本。
* 下载之后使用下面的方法解压。
```
tar -xf paddle_inference.tgz
```
最终会在当前的文件夹中生成`paddle_inference/`的子文件夹。
#### 1.2.2 预测库源码编译
* 如果希望获取最新预测库特性，可以从Paddle github上克隆最新代码，源码编译预测库。
* 可以参考[Paddle预测库官网](https://www.paddlepaddle.org.cn/documentation/docs/zh/advanced_guide/inference_deployment/inference/build_and_install_lib_cn.html)的说明，从github上获取Paddle代码，然后进行编译，生成最新的预测库。使用git获取代码方法如下。
* 可以参考[Paddle预测库官网](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/guides/05_inference_deployment/inference/build_and_install_lib_cn.html)的说明，从github上获取Paddle代码，然后进行编译，生成最新的预测库。使用git获取代码方法如下。
```shell
git clone https://github.com/PaddlePaddle/Paddle.git
......@@ -102,7 +114,7 @@ make -j
make inference_lib_dist
```
更多编译参数选项可以参考Paddle C++预测库官网：[https://www.paddlepaddle.org.cn/documentation/docs/zh/advanced_guide/inference_deployment/inference/build_and_install_lib_cn.html](https://www.paddlepaddle.org.cn/documentation/docs/zh/advanced_guide/inference_deployment/inference/build_and_install_lib_cn.html)。
更多编译参数选项可以参考Paddle C++预测库官网：[https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/guides/05_inference_deployment/inference/build_and_install_lib_cn.html](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/guides/05_inference_deployment/inference/build_and_install_lib_cn.html)。
* 编译完成之后，可以在`build/paddle_inference_install_dir/`文件下看到生成了以下文件及文件夹。
......@@ -117,17 +129,7 @@ build/paddle_inference_install_dir/
其中`paddle`就是C++预测所需的Paddle库，`version.txt`中包含当前预测库的版本信息。
#### 1.2.2 直接下载安装
* [Paddle预测库官网](https://www.paddlepaddle.org.cn/documentation/docs/zh/advanced_guide/inference_deployment/inference/build_and_install_lib_cn.html)上提供了不同cuda版本的Linux预测库，可以在官网查看并选择合适的预测库版本。
* 下载之后使用下面的方法解压。
```
tar -xf paddle_inference.tgz
```
最终会在当前的文件夹中生成`paddle_inference/`的子文件夹。
## 2 开始运行
......@@ -225,7 +227,7 @@ char_list_file ../../ppocr/utils/ppocr_keys_v1.txt # 字典文件
visualize 1 # 是否对结果进行可视化，为1时，会在当前文件夹下保存文件名为`ocr_vis.png`的预测结果。
```
* PaddleOCR也支持多语言的预测，更多细节可以参考[识别文档](../../doc/doc_ch/recognition.md)中的多语言字典与模型部分。
* PaddleOCR也支持多语言的预测，更多支持的语言和模型可以参考[识别文档](../../doc/doc_ch/recognition.md)中的多语言字典与模型部分，如果希望进行多语言预测，只需将修改`tools/config.txt`中的`char_list_file`（字典文件路径）以及`rec_model_dir`（inference模型路径）字段即可。
最终屏幕上会输出检测结果如下。
......@@ -236,4 +238,4 @@ visualize 1 # 是否对结果进行可视化，为1时，会在当前文件夹
### 2.3 注意
* 在使用Paddle预测库时，推荐使用2.0.0-beta0版本的预测库。
* 在使用Paddle预测库时，推荐使用2.0.0版本的预测库。
deploy/cpp_infer/readme_en.md
浏览文件 @ 6ec9aa4d
......@@ -76,10 +76,23 @@ opencv3/
* There are 2 ways to obtain the Paddle inference library, described in detail below.
#### 1.2.1 Direct download and installation
#### 1.2.1 Compile from the source code
* Different cuda versions of the Linux inference library (based on GCC 4.8.2) are provided on the
[Paddle inference library official website](https://www.paddlepaddle.org.cn/documentation/docs/en/develop/guides/05_inference_deployment/inference/build_and_install_lib_en.html). You can view and select the appropriate version of the inference library on the official website.
* After downloading, use the following method to uncompress.
```
tar -xf paddle_inference.tgz
```
Finally you can see the following files in the folder of `paddle_inference/`.
#### 1.2.2 Compile from the source code
* If you want to get the latest Paddle inference library features, you can download the latest code from Paddle github repository and compile the inference library from the source code.
* You can refer to [Paddle inference library] (https://www.paddlepaddle.org.cn/documentation/docs/en/advanced_guide/inference_deployment/inference/build_and_install_lib_en.html) to get the Paddle source code from github, and then compile To generate the latest inference library. The method of using git to access the code is as follows.
* You can refer to [Paddle inference library] (https://www.paddlepaddle.org.cn/documentation/docs/en/develop/guides/05_inference_deployment/inference/build_and_install_lib_en.html) to get the Paddle source code from github, and then compile To generate the latest inference library. The method of using git to access the code is as follows.
```shell
......@@ -106,7 +119,7 @@ make -j
make inference_lib_dist
```
For more compilation parameter options, please refer to the official website of the Paddle C++ inference library:[https://www.paddlepaddle.org.cn/documentation/docs/en/advanced_guide/inference_deployment/inference/build_and_install_lib_en.html](https://www.paddlepaddle.org.cn/documentation/docs/en/advanced_guide/inference_deployment/inference/build_and_install_lib_en.html).
For more compilation parameter options, please refer to the official website of the Paddle C++ inference library:[https://www.paddlepaddle.org.cn/documentation/docs/en/develop/guides/05_inference_deployment/inference/build_and_install_lib_en.html](https://www.paddlepaddle.org.cn/documentation/docs/en/develop/guides/05_inference_deployment/inference/build_and_install_lib_en.html).
* After the compilation process, you can see the following files in the folder of `build/paddle_inference_install_dir/`.
......@@ -122,22 +135,6 @@ build/paddle_inference_install_dir/
Among them, `paddle` is the Paddle library required for C++ prediction later, and `version.txt` contains the version information of the current inference library.
#### 1.2.2 Direct download and installation
* Different cuda versions of the Linux inference library (based on GCC 4.8.2) are provided on the
[Paddle inference library official website](https://www.paddlepaddle.org.cn/documentation/docs/en/advanced_guide/inference_deployment/inference/build_and_install_lib_en.html). You can view and select the appropriate version of the inference library on the official website.
* After downloading, use the following method to uncompress.
```
tar -xf paddle_inference.tgz
```
Finally you can see the following files in the folder of `paddle_inference/`.
## 2. Compile and run the demo
### 2.1 Export the inference model
......@@ -235,7 +232,7 @@ char_list_file ../../ppocr/utils/ppocr_keys_v1.txt # dictionary file
visualize 1 # Whether to visualize the results，when it is set as 1, The prediction result will be save in the image file `./ocr_vis.png`.
```
* Multi-language inference is also supported in PaddleOCR, for more details, please refer to part of multi-language dictionaries and models in [recognition tutorial](../../doc/doc_en/recognition_en.md).
* Multi-language inference is also supported in PaddleOCR, you can refer to [recognition tutorial](../../doc/doc_en/recognition_en.md) for more supported languages and models in PaddleOCR. Specifically, if you want to infer using multi-language models, you just need to modify values of `char_list_file` and `rec_model_dir` in file `tools/config.txt`.
The detection results will be shown on the screen, which is as follows.
......@@ -247,4 +244,4 @@ The detection results will be shown on the screen, which is as follows.
### 2.3 Notes
* Paddle2.0.0-beta0 inference model library is recommended for this toturial.
* Paddle2.0.0 inference model library is recommended for this toturial.
doc/doc_ch/FAQ.md
浏览文件 @ 6ec9aa4d
......@@ -361,13 +361,13 @@
（2）inference模型下载时，如果没有安装wget，可直接点击模型链接或将链接地址复制到浏览器进行下载，并解压放置到相应目录。
#### Q3.1.17：PaddleOCR开源的超轻量模型和通用OCR模型的区别？
**A**：目前PaddleOCR开源了2个中文模型，分别是8.6M超轻量中文模型和通用中文OCR模型。两者对比信息如下：
**A**：目前PaddleOCR开源了2个中文模型，分别是9.4M超轻量中文模型和通用中文OCR模型。两者对比信息如下：
- 相同点：两者使用相同的**算法**和**训练数据**；
- 不同点：不同之处在于**骨干网络**和**通道参数**，超轻量模型使用MobileNetV3作为骨干网络，通用模型使用Resnet50_vd作为检测模型backbone，Resnet34_vd作为识别模型backbone，具体参数差异可对比两种模型训练的配置文件.
|模型|骨干网络|检测训练配置|识别训练配置|
|-|-|-|-|
|8.6M超轻量中文OCR模型|MobileNetV3+MobileNetV3|det_mv3_db.yml|rec_chinese_lite_train.yml|
|9.4M超轻量中文OCR模型|MobileNetV3+MobileNetV3|det_mv3_db.yml|rec_chinese_lite_train.yml|
|通用中文OCR模型|Resnet50_vd+Resnet34_vd|det_r50_vd_db.yml|rec_chinese_common_train.yml|
#### Q3.1.18：如何加入自己的检测算法？
......
doc/doc_ch/angle_class.md
浏览文件 @ 6ec9aa4d
## 文字角度分类
### 方法介绍
文字角度分类主要用于图片非0度的场景下，在这种场景下需要对图片里检测到的文本行进行一个转正的操作。在PaddleOCR系统内，
文字检测之后得到的文本行图片经过仿射变换之后送入识别模型，此时只需要对文字进行一个0和180度的角度分类，因此PaddleOCR内置的
文字角度分类器**只支持了0和180度的分类**。如果想支持更多角度，可以自己修改算法进行支持。
0和180度数据样本例子：
![](../imgs_results/angle_class_example.jpg)
### 数据准备
......@@ -13,7 +21,7 @@ ln -sf <path/to/dataset> <path/to/paddle_ocr>/train_data/cls/dataset
请参考下文组织您的数据。
- 训练集
首先请将训练图片放入同一个文件夹（train_images），并用一个txt文件（cls_gt_train.txt）记录图片路径和标签。
首先建议将训练图片放入同一个文件夹，并用一个txt文件（cls_gt_train.txt）记录图片路径和标签。
**注意：** 默认请将图片路径和图片标签用 `\t` 分割，如用其他方式分割将造成训练报错
......@@ -21,8 +29,8 @@ ln -sf <path/to/dataset> <path/to/paddle_ocr>/train_data/cls/dataset
```
" 图像文件名 图像标注信息 "
train/word_001.jpg 0
train/word_002.jpg 180
train/cls/train/word_001.jpg 0
train/cls/train/word_002.jpg 180
```
最终训练集应有如下文件结构：
......
doc/doc_ch/inference.md
浏览文件 @ 6ec9aa4d
......@@ -141,7 +141,7 @@ python3 tools/infer/predict_det.py --image_dir="./doc/imgs/00018069.jpg" --det_m
![](../imgs_results/det_res_00018069.jpg)
通过参数`limit_type`和`det_limit_side_len`来对图片的尺寸进行限制，
`litmit_type`可选参数为[`max`, `min`]，
`limit_type`可选参数为[`max`, `min`]，
`det_limit_size_len` 为正整数，一般设置为32 的倍数，比如960。
参数默认设置为`limit_type='max', det_limit_side_len=960`。表示网络输入图像的最长边不能超过960，
......
doc/doc_ch/models_list.md
浏览文件 @ 6ec9aa4d
## OCR模型列表（V2.0，2021年1月20日更新）
**说明** ：2.0版模型和[1.1版模型](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_ch/models_list.md)的主要区别在于动态图训练vs.静态图训练，模型性能上无明显差距。
> **说明**
> 1. 2.0版模型和[1.1版模型](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_ch/models_list.md)的主要区别在于动态图训练vs.静态图训练，模型性能上无明显差距。
> 2. 本文档提供的是PPOCR自研模型列表，更多基于公开数据集的算法介绍与预训练模型可以参考：[算法概览文档](./algorithm_overview.md)。
- [一、文本检测模型](#文本检测模型)
- [二、文本识别模型](#文本识别模型)
......@@ -12,9 +16,14 @@ PaddleOCR提供的可下载模型包括`推理模型`、`训练模型`、`预训
|模型类型|模型格式|简介|
|--- | --- | --- |
|推理模型|inference.pdmodel、inference.pdiparams|用于python预测引擎推理，[详情](./inference.md)|
|推理模型|inference.pdmodel、inference.pdiparams|用于预测引擎推理，[详情](./inference.md)|
|训练模型、预训练模型|\*.pdparams、\*.pdopt、\*.states |训练过程中保存的模型的参数、优化器状态和训练中间信息，多用于模型指标评估和恢复训练|
|slim模型|\*.nb|用于lite部署|
|slim模型|\*.nb|经过飞桨模型压缩工具PaddleSlim压缩后的模型，适用于移动端/IoT端等端侧部署场景（需使用飞桨Paddle Lite部署）。|
各个模型的关系如下面的示意图所示。
![](../imgs/model_prod_flow_ch.png)
<a name="文本检测模型"></a>
......
doc/doc_ch/recognition.md
浏览文件 @ 6ec9aa4d
## 文字识别
- [一、数据准备](#数据准备)
- [数据下载](#数据下载)
- [自定义数据集](#自定义数据集)
- [字典](#字典)
- [支持空格](#支持空格)
- [1 数据准备](#数据准备)
- [1.1 自定义数据集](#自定义数据集)
- [1.2 数据下载](#数据下载)
- [1.3 字典](#字典)
- [1.4 支持空格](#支持空格)
- [二、启动训练](#启动训练)
- [1. 数据增强](#数据增强)
- [2. 训练](#训练)
- [3. 小语种](#小语种)
- [2 启动训练](#启动训练)
- [2.1 数据增强](#数据增强)
- [2.2 训练](#训练)
- [2.3 小语种](#小语种)
- [三、评估](#评估)
- [3 评估](#评估)
- [四、预测](#预测)
- [1. 训练引擎预测](#训练引擎预测)
- [4 预测](#预测)
- [4.1 训练引擎预测](#训练引擎预测)
<a name="数据准备"></a>
### 数据准备
### 1. 数据准备
PaddleOCR 支持两种数据格式: `lmdb` 用于训练公开数据，调试算法; `通用数据` 训练自己的数据:
请按如下步骤设置数据集：
PaddleOCR 支持两种数据格式:
- `lmdb` 用于训练以lmdb格式存储的数据集;
- `通用数据` 用于训练以文本文件存储的数据集:
训练数据的默认存储路径是 `PaddleOCR/train_data`,如果您的磁盘上已有数据集，只需创建软链接至数据集目录：
```
# linux and mac os
ln -sf <path/to/dataset> <path/to/paddle_ocr>/train_data/dataset
# windows
mklink /d <path/to/paddle_ocr>/train_data/dataset <path/to/dataset>
```
<a name="数据下载"></a>
* 数据下载
若您本地没有数据集，可以在官网下载 [icdar2015](http://rrc.cvc.uab.es/?ch=4&com=downloads) 数据，用于快速验证。也可以参考[DTRB](https://github.com/clovaai/deep-text-recognition-benchmark#download-lmdb-dataset-for-traininig-and-evaluation-from-here)，下载 benchmark 所需的lmdb格式数据集。
如果希望复现SRN的论文指标，需要下载离线[增广数据](https://pan.baidu.com/s/1-HSZ-ZVdqBF2HaBZ5pRAKA),提取码: y3ry。增广数据是由MJSynth和SynthText做旋转和扰动得到的。数据下载完成后请解压到 {your_path}/PaddleOCR/train_data/data_lmdb_release/training/ 路径下。
<a name="自定义数据集"></a>
* 使用自己数据集
若您希望使用自己的数据进行训练，请参考下文组织您的数据。
<a name="准备数据集"></a>
#### 1.1 自定义数据集
下面以通用数据集为例， 介绍如何准备数据集：
- 训练集
* 训练集
首先请将训练图片放入同一个文件夹（train_images），并用一个txt文件（rec_gt_train.txt）记录图片路径和标签。
建议将训练图片放入同一个文件夹，并用一个txt文件（rec_gt_train.txt）记录图片路径和标签，txt文件里的内容如下:
**注意：** 默认请将图片路径和图片标签用 \t 分割，如用其他方式分割将造成训练报错
**注意：** txt文件中默认请将图片路径和图片标签用 \t 分割，如用其他方式分割将造成训练报错。
```
" 图像文件名 图像标注信息 "
train_data/train_0001.jpg 简单可依赖
train_data/train_0002.jpg 用科技让复杂的世界更简单
```
PaddleOCR 提供了一份用于训练 icdar2015 数据集的标签文件，通过以下方式下载：
```
# 训练集标签
wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_train.txt
# 测试集标签
wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_test.txt
```
PaddleOCR 也提供了数据格式转换脚本，可以将官网 label 转换支持的数据格式。 数据转换工具在 `ppocr/utils/gen_label.py`, 这里以训练集为例：
```
# 将官网下载的标签文件转换为 rec_gt_label.txt
python gen_label.py --mode="rec" --input_path="{path/of/origin/label}" --output_label="rec_gt_label.txt"
train_data/rec/train/word_001.jpg 简单可依赖
train_data/rec/train/word_002.jpg 用科技让复杂的世界更简单
...
```
最终训练集应有如下文件结构：
```
|-train_data
|-ic15_data
|-rec
|- rec_gt_train.txt
|- train
|- word_001.png
......@@ -89,7 +71,7 @@ python gen_label.py --mode="rec" --input_path="{path/of/origin/label}" --output_
```
|-train_data
|-ic15_data
|-rec
|- rec_gt_test.txt
|- test
|- word_001.jpg
......@@ -97,8 +79,33 @@ python gen_label.py --mode="rec" --input_path="{path/of/origin/label}" --output_
|- word_003.jpg
| ...
```
<a name="数据下载"></a>
1.2 数据下载
若您本地没有数据集，可以在官网下载 [icdar2015](http://rrc.cvc.uab.es/?ch=4&com=downloads) 数据，用于快速验证。也可以参考[DTRB](https://github.com/clovaai/deep-text-recognition-benchmark#download-lmdb-dataset-for-traininig-and-evaluation-from-here) ，下载 benchmark 所需的lmdb格式数据集。
如果你使用的是icdar2015的公开数据集，PaddleOCR 提供了一份用于训练 icdar2015 数据集的标签文件，通过以下方式下载：
如果希望复现SRN的论文指标，需要下载离线[增广数据](https://pan.baidu.com/s/1-HSZ-ZVdqBF2HaBZ5pRAKA),提取码: y3ry。增广数据是由MJSynth和SynthText做旋转和扰动得到的。数据下载完成后请解压到 {your_path}/PaddleOCR/train_data/data_lmdb_release/training/ 路径下。
```
# 训练集标签
wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_train.txt
# 测试集标签
wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_test.txt
```
PaddleOCR 也提供了数据格式转换脚本，可以将官网 label 转换支持的数据格式。 数据转换工具在 `ppocr/utils/gen_label.py`, 这里以训练集为例：
```
# 将官网下载的标签文件转换为 rec_gt_label.txt
python gen_label.py --mode="rec" --input_path="{path/of/origin/label}" --output_label="rec_gt_label.txt"
```
<a name="字典"></a>
- 字典
1.3 字典
最后需要提供一个字典（{word_dict_name}.txt），使模型在训练时，可以将所有出现的字符映射为字典的索引。
......@@ -115,6 +122,10 @@ n
word_dict.txt 每行有一个单字，将字符与数字索引映射在一起，“and” 将被映射成 [2 5 1]
* 内置字典
PaddleOCR内置了一部分字典，可以按需使用。
`ppocr/utils/ppocr_keys_v1.txt` 是一个包含6623个字符的中文字典
`ppocr/utils/ic15_dict.txt` 是一个包含36个字符的英文字典
......@@ -130,7 +141,7 @@ word_dict.txt 每行有一个单字，将字符与数字索引映射在一起，
`ppocr/utils/dict/en_dict.txt` 是一个包含63个字符的英文字典
您可以按需使用。
目前的多语言模型仍处在demo阶段，会持续优化模型并补充语种，**非常欢迎您为我们提供其他语言的字典和字体**，
如您愿意可将字典文件提交至 [dict](../../ppocr/utils/dict)，我们会在Repo中感谢您。
......@@ -141,13 +152,13 @@ word_dict.txt 每行有一个单字，将字符与数字索引映射在一起，
并将 `character_type` 设置为 `ch`。
<a name="支持空格"></a>
- 添加空格类别
1.4 添加空格类别
如果希望支持识别"空格"类别, 请将yml文件中的 `use_space_char` 字段设置为 `True`。
<a name="启动训练"></a>
### 启动训练
### 2. 启动训练
PaddleOCR提供了训练脚本、评估脚本和预测脚本，本节将以 CRNN 识别模型为例：
......@@ -172,7 +183,7 @@ tar -xf rec_mv3_none_bilstm_ctc_v2.0_train.tar && rm -rf rec_mv3_none_bilstm_ctc
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/rec/rec_icdar15_train.yml
```
<a name="数据增强"></a>
- 数据增强
#### 2.1 数据增强
PaddleOCR提供了多种数据增强方式，如果您希望在训练时加入扰动，请在配置文件中设置 `distort: true`。
......@@ -183,7 +194,7 @@ PaddleOCR提供了多种数据增强方式，如果您希望在训练时加入
*由于OpenCV的兼容性问题，扰动操作暂时只支持Linux*
<a name="训练"></a>
- 训练
#### 2.2 训练
PaddleOCR支持训练和评估交替进行, 可以在 `configs/rec/rec_icdar15_train.yml` 中修改 `eval_batch_step` 设置评估频率，默认每500个iter评估一次。评估过程中默认将最佳acc模型，保存为 `output/rec_CRNN/best_accuracy` 。
......@@ -272,7 +283,7 @@ Eval:
**注意，预测/评估时的配置文件请务必与训练一致。**
<a name="小语种"></a>
- 小语种
#### 2.3 小语种
PaddleOCR目前已支持26种（除中文外）语种识别，`configs/rec/multi_languages` 路径下提供了一个多语言的配置文件模版: [rec_multi_language_lite_train.yml](../../configs/rec/multi_language/rec_multi_language_lite_train.yml)。
......@@ -415,7 +426,7 @@ Eval:
...
```
<a name="评估"></a>
### 评估
### 3 评估
评估数据集可以通过 `configs/rec/rec_icdar15_train.yml` 修改Eval中的 `label_file_path` 设置。
......@@ -425,10 +436,10 @@ python3 -m paddle.distributed.launch --gpus '0' tools/eval.py -c configs/rec/rec
```
<a name="预测"></a>
### 预测
### 4 预测
<a name="训练引擎预测"></a>
* 训练引擎的预测
#### 4.1 训练引擎的预测
使用 PaddleOCR 训练好的模型，可以通过以下脚本进行快速预测。
......
doc/doc_ch/whl.md
浏览文件 @ 6ec9aa4d
# paddleocr package使用说明
## 快速上手
## 1 快速上手
### 安装whl包
### 1.1 安装whl包
pip安装
```bash
......@@ -14,9 +14,12 @@ pip install "paddleocr>=2.0.1" # 推荐使用2.0.1+版本
python3 setup.py bdist_wheel
pip3 install dist/paddleocr-x.x.x-py3-none-any.whl # x.x.x是paddleocr的版本号
```
### 1. 代码使用
* 检测+分类+识别全流程
## 2 使用
### 2.1 代码使用
paddleocr whl包会自动下载ppocr轻量级模型作为默认模型，可以根据第3节**自定义模型**进行自定义更换。
* 检测+方向分类器+识别全流程
```python
from paddleocr import PaddleOCR, draw_ocr
# Paddleocr目前支持中英文、英文、法语、德语、韩语、日语，可以通过修改lang参数进行切换
......@@ -84,7 +87,7 @@ im_show.save('result.jpg')
</div>
* 分类+识别
* 方向分类器+识别
```python
from paddleocr import PaddleOCR
ocr = PaddleOCR(use_angle_cls=True) # need to run only once to download and load model into memory
......@@ -143,7 +146,7 @@ for line in result:
['韩国小馆', 0.9907421]
```
* 单独执行分类
* 单独执行方向分类器
```python
from paddleocr import PaddleOCR
ocr = PaddleOCR(use_angle_cls=True) # need to run only once to download and load model into memory
......@@ -157,14 +160,14 @@ for line in result:
['0', 0.9999924]
```
### 通过命令行使用
### 2.2 通过命令行使用
查看帮助信息
```bash
paddleocr -h
```
* 检测+分类+识别全流程
* 检测+方向分类器+识别全流程
```bash
paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg --use_angle_cls true
```
......@@ -188,7 +191,7 @@ paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg
......
```
* 分类+识别
* 方向分类器+识别
```bash
paddleocr --image_dir PaddleOCR/doc/imgs_words/ch/word_1.jpg --use_angle_cls true --det false
```
......@@ -220,7 +223,7 @@ paddleocr --image_dir PaddleOCR/doc/imgs_words/ch/word_1.jpg --det false
['韩国小馆', 0.9907421]
```
* 单独执行分类
* 单独执行方向分类器
```bash
paddleocr --image_dir PaddleOCR/doc/imgs_words/ch/word_1.jpg --use_angle_cls true --det false --rec false
```
......@@ -230,11 +233,11 @@ paddleocr --image_dir PaddleOCR/doc/imgs_words/ch/word_1.jpg --use_angle_cls tru
['0', 0.9999924]
```
## 自定义模型
## 3 自定义模型
当内置模型无法满足需求时，需要使用到自己训练的模型。
首先，参照[inference.md](./inference.md) 第一节转换将检测、分类和识别模型转换为inference模型，然后按照如下方式使用
### 代码使用
### 3.1 代码使用
```python
from paddleocr import PaddleOCR, draw_ocr
# 模型路径下必须含有model和params文件
......@@ -255,17 +258,17 @@ im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```
### 通过命令行使用
### 3.2 通过命令行使用
```bash
paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg --det_model_dir {your_det_model_dir} --rec_model_dir {your_rec_model_dir} --rec_char_dict_path {your_rec_char_dict_path} --cls_model_dir {your_cls_model_dir} --use_angle_cls true
```
### 使用网络图片或者numpy数组作为输入
## 4 使用网络图片或者numpy数组作为输入
1. 网络图片
### 4.1 网络图片
代码使用
- 代码使用
```python
from paddleocr import PaddleOCR, draw_ocr
# Paddleocr目前支持中英文、英文、法语、德语、韩语、日语，可以通过修改lang参数进行切换
......@@ -286,12 +289,12 @@ im_show = draw_ocr(image, boxes, txts, scores, font_path='/path/to/PaddleOCR/doc
im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```
命令行模式
- 命令行模式
```bash
paddleocr --image_dir http://n.sinaimg.cn/ent/transform/w630h933/20171222/o111-fypvuqf1838418.jpg --use_angle_cls=true
```
2. numpy数组
### 4.2 numpy数组
仅通过代码使用时支持numpy数组作为输入
```python
from paddleocr import PaddleOCR, draw_ocr
......@@ -301,7 +304,7 @@ ocr = PaddleOCR(use_angle_cls=True, lang="ch") # need to run only once to downlo
img_path = 'PaddleOCR/doc/imgs/11.jpg'
img = cv2.imread(img_path)
# img = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY), 如果你自己训练的模型支持灰度图，可以将这句话的注释取消
result = ocr.ocr(img_path, cls=True)
result = ocr.ocr(img, cls=True)
for line in result:
print(line)
......@@ -316,7 +319,7 @@ im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```
## 参数说明
## 5 参数说明
| 字段 | 说明 | 默认值 |
|-------------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-------------------------|
......
doc/doc_en/angle_class_en.md
浏览文件 @ 6ec9aa4d
## TEXT ANGLE CLASSIFICATION
### Method introduction
The angle classification is used in the scene where the image is not 0 degrees. In this scene, it is necessary to perform a correction operation on the text line detected in the picture. In the PaddleOCR system,
The text line image obtained after text detection is sent to the recognition model after affine transformation. At this time, only a 0 and 180 degree angle classification of the text is required, so the built-in PaddleOCR text angle classifier **only supports 0 and 180 degree classification**. If you want to support more angles, you can modify the algorithm yourself to support.
Example of 0 and 180 degree data samples：
![](../imgs_results/angle_class_example.jpg)
### DATA PREPARATION
Please organize the dataset as follows:
......
doc/doc_en/inference_en.md
浏览文件 @ 6ec9aa4d
......@@ -148,7 +148,7 @@ The visual text detection results are saved to the ./inference_results folder by
![](../imgs_results/det_res_00018069.jpg)
You can use the parameters `limit_type` and `det_limit_side_len` to limit the size of the input image,
The optional parameters of `litmit_type` are [`max`, `min`], and
The optional parameters of `limit_type` are [`max`, `min`], and
`det_limit_size_len` is a positive integer, generally set to a multiple of 32, such as 960.
The default setting of the parameters is `limit_type='max', det_limit_side_len=960`. Indicates that the longest side of the network input image cannot exceed 960,
......
doc/doc_en/models_list_en.md
浏览文件 @ 6ec9aa4d
## OCR model list（V2.0, updated on 2021.1.20）
**Note** : Compared with [models 1.1](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_en/models_list_en.md), which are trained with static graph programming paradigm, models 2.0 are the dynamic graph trained version and achieve close performance.
> **Note**
> 1. Compared with [models 1.1](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_en/models_list_en.md), which are trained with static graph programming paradigm, models 2.0 are the dynamic graph trained version and achieve close performance.
> 2. All models in this tutorial are all ppocr-series models, for more introduction of algorithms and models based on public dataset, you can refer to [algorithm overview tutorial](./algorithm_overview_en.md).
- [1. Text Detection Model](#Detection)
- [2. Text Recognition Model](#Recognition)
......@@ -12,9 +14,13 @@ The downloadable models provided by PaddleOCR include `inference model`, `traine
|model type|model format|description|
|--- | --- | --- |
|inference model|inference.pdmodel、inference.pdiparams|Used for reasoning based on Python prediction engine，[detail](./inference_en.md)|
|inference model|inference.pdmodel、inference.pdiparams|Used for inference based on Paddle inference engine，[detail](./inference_en.md)|
|trained model, pre-trained model|\*.pdparams、\*.pdopt、\*.states |The checkpoints model saved in the training process, which stores the parameters of the model, mostly used for model evaluation and continuous training.|
|slim model|\*.nb|Generally used for Lite deployment|
|slim model|\*.nb| Model compressed by PaddleSim (a model compression tool using PaddlePaddle), which is suitable for mobile-side deployment scenarios (Paddle-Lite is needed for slim model deployment). |
Relationship of the above models is as follows.
![](../imgs_en/model_prod_flow_en.png)
<a name="Detection"></a>
### 1. Text Detection Model
......
doc/doc_en/recognition_en.md
浏览文件 @ 6ec9aa4d
## TEXT RECOGNITION
- [DATA PREPARATION](#DATA_PREPARATION)
- [Dataset Download](#Dataset_download)
- [Costom Dataset](#Costom_Dataset)
- [Dictionary](#Dictionary)
- [Add Space Category](#Add_space_category)
- [1 DATA PREPARATION](#DATA_PREPARATION)
- [1.1 Costom Dataset](#Costom_Dataset)
- [1.2 Dataset Download](#Dataset_download)
- [1.3 Dictionary](#Dictionary)
- [1.4 Add Space Category](#Add_space_category)
- [TRAINING](#TRAINING)
- [Data Augmentation](#Data_Augmentation)
- [Training](#Training)
- [Multi-language](#Multi_language)
- [2 TRAINING](#TRAINING)
- [2.1 Data Augmentation](#Data_Augmentation)
- [2.2 Training](#Training)
- [2.3 Multi-language](#Multi_language)
- [EVALUATION](#EVALUATION)
- [3 EVALUATION](#EVALUATION)
- [PREDICTION](#PREDICTION)
- [Training engine prediction](#Training_engine_prediction)
- [4 PREDICTION](#PREDICTION)
- [4.1 Training engine prediction](#Training_engine_prediction)
<a name="DATA_PREPARATION"></a>
### DATA PREPARATION
PaddleOCR supports two data formats: `LMDB` is used to train public data and evaluation algorithms; `general data` is used to train your own data:
PaddleOCR supports two data formats:
- `LMDB` is used to train data sets stored in lmdb format;
- `general data` is used to train data sets stored in text files:
Please organize the dataset as follows:
The default storage path for training data is `PaddleOCR/train_data`, if you already have a dataset on your disk, just create a soft link to the dataset directory:
```
# linux and mac os
ln -sf <path/to/dataset> <path/to/paddle_ocr>/train_data/dataset
# windows
mklink /d <path/to/paddle_ocr>/train_data/dataset <path/to/dataset>
```
<a name="Dataset_download"></a>
* Dataset download
If you do not have a dataset locally, you can download it on the official website [icdar2015](http://rrc.cvc.uab.es/?ch=4&com=downloads). Also refer to [DTRB](https://github.com/clovaai/deep-text-recognition-benchmark#download-lmdb-dataset-for-traininig-and-evaluation-from-here)，download the lmdb format dataset required for benchmark
If you want to reproduce the paper indicators of SRN, you need to download offline [augmented data](https://pan.baidu.com/s/1-HSZ-ZVdqBF2HaBZ5pRAKA), extraction code: y3ry. The augmented data is obtained by rotation and perturbation of mjsynth and synthtext. Please unzip the data to {your_path}/PaddleOCR/train_data/data_lmdb_Release/training/path.
<a name="Costom_Dataset"></a>
* Use your own dataset:
#### 1.1 Costom dataset
If you want to use your own data for training, please refer to the following to organize your data.
- Training set
First put the training images in the same folder (train_images), and use a txt file (rec_gt_train.txt) to store the image path and label.
It is recommended to put the training images in the same folder, and use a txt file (rec_gt_train.txt) to store the image path and label. The contents of the txt file are as follows:
* Note: by default, the image path and image label are split with \t, if you use other methods to split, it will cause training error
```
" Image file name Image annotation "
train_data/train_0001.jpg 简单可依赖
train_data/train_0002.jpg 用科技让复杂的世界更简单
```
PaddleOCR provides label files for training the icdar2015 dataset, which can be downloaded in the following ways:
```
# Training set label
wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_train.txt
# Test Set Label
wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_test.txt
train_data/rec/train/word_001.jpg 简单可依赖
train_data/rec/train/word_002.jpg 用科技让复杂的世界更简单
...
```
The final training set should have the following file structure:
```
|-train_data
|-ic15_data
|-rec
|- rec_gt_train.txt
|- train
|- word_001.png
......@@ -82,6 +73,7 @@ Similar to the training set, the test set also needs to be provided a folder con
```
|-train_data
|-rec
|-ic15_data
|- rec_gt_test.txt
|- test
......@@ -90,8 +82,25 @@ Similar to the training set, the test set also needs to be provided a folder con
|- word_003.jpg
| ...
```
<a name="Dataset_download"></a>
#### 1.2 Dataset download
If you do not have a dataset locally, you can download it on the official website [icdar2015](http://rrc.cvc.uab.es/?ch=4&com=downloads). Also refer to [DTRB](https://github.com/clovaai/deep-text-recognition-benchmark#download-lmdb-dataset-for-traininig-and-evaluation-from-here) ，download the lmdb format dataset required for benchmark
If you want to reproduce the paper indicators of SRN, you need to download offline [augmented data](https://pan.baidu.com/s/1-HSZ-ZVdqBF2HaBZ5pRAKA), extraction code: y3ry. The augmented data is obtained by rotation and perturbation of mjsynth and synthtext. Please unzip the data to {your_path}/PaddleOCR/train_data/data_lmdb_Release/training/path.
PaddleOCR provides label files for training the icdar2015 dataset, which can be downloaded in the following ways:
```
# Training set label
wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_train.txt
# Test Set Label
wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_test.txt
```
<a name="Dictionary"></a>
- Dictionary
#### 1.3 Dictionary
Finally, a dictionary ({word_dict_name}.txt) needs to be provided so that when the model is trained, all the characters that appear can be mapped to the dictionary index.
......@@ -108,6 +117,8 @@ n
In `word_dict.txt`, there is a single word in each line, which maps characters and numeric indexes together, e.g "and" will be mapped to [2 5 1]
PaddleOCR has built-in dictionaries, which can be used on demand.
`ppocr/utils/ppocr_keys_v1.txt` is a Chinese dictionary with 6623 characters.
`ppocr/utils/ic15_dict.txt` is an English dictionary with 63 characters
......@@ -123,8 +134,6 @@ In `word_dict.txt`, there is a single word in each line, which maps characters a
`ppocr/utils/dict/en_dict.txt` is a English dictionary with 63 characters
You can use it on demand.
The current multi-language model is still in the demo stage and will continue to optimize the model and add languages. **You are very welcome to provide us with dictionaries and fonts in other languages**,
If you like, you can submit the dictionary file to [dict](../../ppocr/utils/dict) and we will thank you in the Repo.
......@@ -136,14 +145,14 @@ To customize the dict file, please modify the `character_dict_path` field in `co
If you need to customize dic file, please add character_dict_path field in configs/rec/rec_icdar15_train.yml to point to your dictionary path. And set character_type to ch.
<a name="Add_space_category"></a>
- Add space category
#### 1.4 Add space category
If you want to support the recognition of the `space` category, please set the `use_space_char` field in the yml file to `True`.
**Note: use_space_char only takes effect when character_type=ch**
<a name="TRAINING"></a>
### TRAINING
### 2 TRAINING
PaddleOCR provides training scripts, evaluation scripts, and prediction scripts. In this section, the CRNN recognition model will be used as an example:
......@@ -166,7 +175,7 @@ Start training:
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/rec/rec_icdar15_train.yml
```
<a name="Data_Augmentation"></a>
- Data Augmentation
#### 2.1 Data Augmentation
PaddleOCR provides a variety of data augmentation methods. If you want to add disturbance during training, please set `distort: true` in the configuration file.
......@@ -175,7 +184,7 @@ The default perturbation methods are: cvtColor, blur, jitter, Gasuss noise, rand
Each disturbance method is selected with a 50% probability during the training process. For specific code implementation, please refer to: [img_tools.py](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/ppocr/data/rec/img_tools.py)
<a name="Training"></a>
- Training
#### 2.2 Training
PaddleOCR supports alternating training and evaluation. You can modify `eval_batch_step` in `configs/rec/rec_icdar15_train.yml` to set the evaluation frequency. By default, it is evaluated every 500 iter and the best acc model is saved under `output/rec_CRNN/best_accuracy` during the evaluation process.
......@@ -268,7 +277,7 @@ Eval:
**Note that the configuration file for prediction/evaluation must be consistent with the training.**
<a name="Multi_language"></a>
- Multi-language
#### 2.3 Multi-language
PaddleOCR currently supports 26 (except Chinese) language recognition. A multi-language configuration file template is
provided under the path `configs/rec/multi_languages`: [rec_multi_language_lite_train.yml](../../configs/rec/multi_language/rec_multi_language_lite_train.yml)。
......@@ -420,7 +429,7 @@ Eval:
```
<a name="EVALUATION"></a>
### EVALUATION
### 3 EVALUATION
The evaluation dataset can be set by modifying the `Eval.dataset.label_file_list` field in the `configs/rec/rec_icdar15_train.yml` file.
......@@ -430,10 +439,10 @@ python3 -m paddle.distributed.launch --gpus '0' tools/eval.py -c configs/rec/rec
```
<a name="PREDICTION"></a>
### PREDICTION
### 4 PREDICTION
<a name="Training_engine_prediction"></a>
* Training engine prediction
#### 4.1 Training engine prediction
Using the model trained by paddleocr, you can quickly get prediction through the following script.
......
doc/doc_en/whl_en.md
浏览文件 @ 6ec9aa4d
# paddleocr package
## Get started quickly
### install package
## 1 Get started quickly
### 1.1 install package
install by pypi
```bash
pip install "paddleocr>=2.0.1" # Recommend to use version 2.0.1+
......@@ -12,9 +12,11 @@ build own whl package and install
python3 setup.py bdist_wheel
pip3 install dist/paddleocr-x.x.x-py3-none-any.whl # x.x.x is the version of paddleocr
```
### 1. Use by code
## 2 Use
### 2.1 Use by code
The paddleocr whl package will automatically download the ppocr lightweight model as the default model, which can be customized and replaced according to the section 3 **Custom Model**.
* detection classification and recognition
* detection angle classification and recognition
```python
from paddleocr import PaddleOCR,draw_ocr
# Paddleocr supports Chinese, English, French, German, Korean and Japanese.
......@@ -163,7 +165,7 @@ Output will be a list, each item contains classification result and confidence
['0', 0.99999964]
```
### Use by command line
### 2.2 Use by command line
show help information
```bash
......@@ -239,11 +241,11 @@ Output will be a list, each item contains classification result and confidence
['0', 0.99999964]
```
## Use custom model
## 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
### 1. Use by code
### 3.1 Use by code
```python
from paddleocr import PaddleOCR,draw_ocr
......@@ -265,17 +267,17 @@ im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```
### Use by command line
### 3.2 Use by command line
```bash
paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg --det_model_dir {your_det_model_dir} --rec_model_dir {your_rec_model_dir} --rec_char_dict_path {your_rec_char_dict_path} --cls_model_dir {your_cls_model_dir} --use_angle_cls true
```
### Use web images or numpy array as input
## 4 Use web images or numpy array as input
1. Web image
### 4.1 Web image
Use by code
- Use by code
```python
from paddleocr import PaddleOCR, draw_ocr
ocr = PaddleOCR(use_angle_cls=True, lang="ch") # need to run only once to download and load model into memory
......@@ -294,12 +296,12 @@ im_show = draw_ocr(image, boxes, txts, scores, font_path='/path/to/PaddleOCR/doc
im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```
Use by command line
- Use by command line
```bash
paddleocr --image_dir http://n.sinaimg.cn/ent/transform/w630h933/20171222/o111-fypvuqf1838418.jpg --use_angle_cls=true
```
2. Numpy array
### 4.2 Numpy array
Support numpy array as input only when used by code
```python
......@@ -324,7 +326,7 @@ im_show.save('result.jpg')
```
## Parameter Description
## 5 Parameter Description
| Parameter | Description | Default value |
|-------------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-------------------------|
......
doc/joinus.PNG
查看替换文件 @ df1c97af 浏览文件 @ 6ec9aa4d
doc/joinus.PNG

114.4 KB | W: | H:

doc/joinus.PNG

109.2 KB | W: | H:

doc/joinus.PNG
doc/joinus.PNG
doc/joinus.PNG
doc/joinus.PNG
  • 2-up
  • Swipe
  • Onion skin
ppocr/data/imaug/label_ops.py
浏览文件 @ 6ec9aa4d
......@@ -215,7 +215,7 @@ class AttnLabelEncode(BaseRecLabelEncode):
return None
data['length'] = np.array(len(text))
text = [0] + text + [len(self.character) - 1] + [0] * (self.max_text_len
- len(text) - 1)
- len(text) - 2)
data['label'] = np.array(text)
return data
......@@ -255,13 +255,13 @@ class SRNLabelEncode(BaseRecLabelEncode):
def __call__(self, data):
text = data['label']
text = self.encode(text)
char_num = len(self.character_str)
char_num = len(self.character)
if text is None:
return None
if len(text) > self.max_text_len:
return None
data['length'] = np.array(len(text))
text = text + [char_num] * (self.max_text_len - len(text))
text = text + [char_num - 1] * (self.max_text_len - len(text))
data['label'] = np.array(text)
return data
......
ppocr/data/imaug/make_shrink_map.py
浏览文件 @ 6ec9aa4d
......@@ -84,11 +84,12 @@ class MakeShrinkMap(object):
return polygons, ignore_tags
def polygon_area(self, polygon):
# return cv2.contourArea(polygon.astype(np.float32))
edge = 0
for i in range(polygon.shape[0]):
next_index = (i + 1) % polygon.shape[0]
edge += (polygon[next_index, 0] - polygon[i, 0]) * (
polygon[next_index, 1] - polygon[i, 1])
return edge / 2.
"""
compute polygon area
"""
area = 0
q = polygon[-1]
for p in polygon:
area += p[0] * q[1] - p[1] * q[0]
q = p
return area / 2.0
ppocr/data/imaug/operators.py
浏览文件 @ 6ec9aa4d
......@@ -185,8 +185,8 @@ class DetResizeForTest(object):
resize_h = int(h * ratio)
resize_w = int(w * ratio)
resize_h = int(round(resize_h / 32) * 32)
resize_w = int(round(resize_w / 32) * 32)
resize_h = max(int(round(resize_h / 32) * 32), 32)
resize_w = max(int(round(resize_w / 32) * 32), 32)
try:
if int(resize_w) <= 0 or int(resize_h) <= 0:
......
ppocr/metrics/rec_metric.py
浏览文件 @ 6ec9aa4d
......@@ -29,7 +29,7 @@ class RecMetric(object):
pred = pred.replace(" ", "")
target = target.replace(" ", "")
norm_edit_dis += Levenshtein.distance(pred, target) / max(
len(pred), len(target))
len(pred), len(target), 1)
if pred == target:
correct_num += 1
all_num += 1
......
ppocr/modeling/heads/rec_att_head.py
浏览文件 @ 6ec9aa4d
......@@ -146,6 +146,9 @@ class AttentionLSTM(nn.Layer):
else:
targets = paddle.zeros(shape=[batch_size], dtype="int32")
probs = None
char_onehots = None
outputs = None
alpha = None
for i in range(num_steps):
char_onehots = self._char_to_onehot(
......
tools/eval.py
浏览文件 @ 6ec9aa4d
......@@ -47,6 +47,7 @@ def main():
config['Architecture']["Head"]['out_channels'] = len(
getattr(post_process_class, 'character'))
model = build_model(config['Architecture'])
use_srn = config['Architecture']['algorithm'] == "SRN"
best_model_dict = init_model(config, model, logger)
if len(best_model_dict):
......@@ -59,7 +60,7 @@ def main():
# start eval
metirc = program.eval(model, valid_dataloader, post_process_class,
eval_class)
eval_class, use_srn)
logger.info('metric eval ***************')
for k, v in metirc.items():
logger.info('{}:{}'.format(k, v))
......
tools/infer/predict_rec.py
浏览文件 @ 6ec9aa4d
......@@ -54,6 +54,13 @@ class TextRecognizer(object):
"character_dict_path": args.rec_char_dict_path,
"use_space_char": args.use_space_char
}
elif self.rec_algorithm == "RARE":
postprocess_params = {
'name': 'AttnLabelDecode',
"character_type": args.rec_char_type,
"character_dict_path": args.rec_char_dict_path,
"use_space_char": args.use_space_char
}
self.postprocess_op = build_post_process(postprocess_params)
self.predictor, self.input_tensor, self.output_tensors = \
utility.create_predictor(args, 'rec', logger)
......
tools/program.py
浏览文件 @ 6ec9aa4d
......@@ -163,6 +163,11 @@ def train(config,
if type(eval_batch_step) == list and len(eval_batch_step) >= 2:
start_eval_step = eval_batch_step[0]
eval_batch_step = eval_batch_step[1]
if len(valid_dataloader) == 0:
logger.info(
'No Images in eval dataset, evaluation during training will be disabled'
)
start_eval_step = 1e111
logger.info(
"During the training process, after the {}th iteration, an evaluation is run every {} iterations".
format(start_eval_step, eval_batch_step))
......@@ -177,6 +182,8 @@ def train(config,
model_average = False
model.train()
use_srn = config['Architecture']['algorithm'] == "SRN"
if 'start_epoch' in best_model_dict:
start_epoch = best_model_dict['start_epoch']
else:
......@@ -195,7 +202,7 @@ def train(config,
break
lr = optimizer.get_lr()
images = batch[0]
if config['Architecture']['algorithm'] == "SRN":
if use_srn:
others = batch[-4:]
preds = model(images, others)
model_average = True
......@@ -251,8 +258,12 @@ def train(config,
min_average_window=10000,
max_average_window=15625)
Model_Average.apply()
cur_metric = eval(model, valid_dataloader, post_process_class,
eval_class)
cur_metric = eval(
model,
valid_dataloader,
post_process_class,
eval_class,
use_srn=use_srn)
cur_metric_str = 'cur metric, {}'.format(', '.join(
['{}: {}'.format(k, v) for k, v in cur_metric.items()]))
logger.info(cur_metric_str)
......@@ -316,7 +327,8 @@ def train(config,
return
def eval(model, valid_dataloader, post_process_class, eval_class):
def eval(model, valid_dataloader, post_process_class, eval_class,
use_srn=False):
model.eval()
with paddle.no_grad():
total_frame = 0.0
......@@ -327,7 +339,8 @@ def eval(model, valid_dataloader, post_process_class, eval_class):
break
images = batch[0]
start = time.time()
if "SRN" in str(model.head):
if use_srn:
others = batch[-4:]
preds = model(images, others)
else:
......
tools/train.py
浏览文件 @ 6ec9aa4d
......@@ -50,6 +50,12 @@ def main(config, device, logger, vdl_writer):
# build dataloader
train_dataloader = build_dataloader(config, 'Train', device, logger)
if len(train_dataloader) == 0:
logger.error(
'No Images in train dataset, please check annotation file and path in the configuration file'
)
return
if config['Eval']:
valid_dataloader = build_dataloader(config, 'Eval', device, logger)
else:
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或