Merge remote-tracking branch 'Evezerest/dygraph' into dygraph

PPOCRLabel/PPOCRLabel.py

@@ -152,16 +152,6 @@ class MainWindow(QMainWindow):
         self.fileListWidget.setIconSize(QSize(25, 25))
         filelistLayout.addWidget(self.fileListWidget)
 
-        self.AutoRecognition = QToolButton()
-        self.AutoRecognition.setToolButtonStyle(Qt.ToolButtonTextBesideIcon)
-        self.AutoRecognition.setIcon(newIcon('Auto'))
-        autoRecLayout = QHBoxLayout()
-        autoRecLayout.setContentsMargins(0, 0, 0, 0)
-        autoRecLayout.addWidget(self.AutoRecognition)
-        autoRecContainer = QWidget()
-        autoRecContainer.setLayout(autoRecLayout)
-        filelistLayout.addWidget(autoRecContainer)
-
         fileListContainer = QWidget()
         fileListContainer.setLayout(filelistLayout)
         self.fileListName = getStr('fileList')
@@ -172,17 +162,30 @@ class MainWindow(QMainWindow):
 
         #  ================== Key List  ==================
         if self.kie_mode:
-            # self.keyList = QListWidget()
             self.keyList = UniqueLabelQListWidget()
-            # self.keyList.itemSelectionChanged.connect(self.keyListSelectionChanged)
-            # self.keyList.itemDoubleClicked.connect(self.editBox)
-            # self.keyList.itemChanged.connect(self.keyListItemChanged)
+
+            # set key list height
+            key_list_height = int(QApplication.desktop().height() // 4)
+            if key_list_height < 50:
+                key_list_height = 50
+            self.keyList.setMaximumHeight(key_list_height)
+
             self.keyListDockName = getStr('keyListTitle')
             self.keyListDock = QDockWidget(self.keyListDockName, self)
             self.keyListDock.setWidget(self.keyList)
             self.keyListDock.setFeatures(QDockWidget.NoDockWidgetFeatures)
             filelistLayout.addWidget(self.keyListDock)
 
+        self.AutoRecognition = QToolButton()
+        self.AutoRecognition.setToolButtonStyle(Qt.ToolButtonTextBesideIcon)
+        self.AutoRecognition.setIcon(newIcon('Auto'))
+        autoRecLayout = QHBoxLayout()
+        autoRecLayout.setContentsMargins(0, 0, 0, 0)
+        autoRecLayout.addWidget(self.AutoRecognition)
+        autoRecContainer = QWidget()
+        autoRecContainer.setLayout(autoRecLayout)
+        filelistLayout.addWidget(autoRecContainer)
+
         #  ================== Right Area  ==================
         listLayout = QVBoxLayout()
         listLayout.setContentsMargins(0, 0, 0, 0)
@@ -431,8 +434,7 @@ class MainWindow(QMainWindow):
         #  ================== New Actions ==================
 
         edit = action(getStr('editLabel'), self.editLabel,
-                      'Ctrl+E', 'edit', getStr('editLabelDetail'),
-                      enabled=False)
+                      'Ctrl+E', 'edit', getStr('editLabelDetail'), enabled=False)
 
         AutoRec = action(getStr('autoRecognition'), self.autoRecognition,
                          '', 'Auto', getStr('autoRecognition'), enabled=False)
@@ -465,11 +467,10 @@ class MainWindow(QMainWindow):
                       'Ctrl+Z', "undo", getStr("undo"), enabled=False)
 
         change_cls = action(getStr("keyChange"), self.change_box_key,
-                            'Ctrl+B', "edit", getStr("keyChange"), enabled=False)
+                            'Ctrl+X', "edit", getStr("keyChange"), enabled=False)
 
         lock = action(getStr("lockBox"), self.lockSelectedShape,
-                      None, "lock", getStr("lockBoxDetail"),
-                      enabled=False)
+                      None, "lock", getStr("lockBoxDetail"), enabled=False)
 
         self.editButton.setDefaultAction(edit)
         self.newButton.setDefaultAction(create)
@@ -534,7 +535,8 @@ class MainWindow(QMainWindow):
                               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, lock),
+                                        None, rotateLeft, rotateRight, None, color1, self.drawSquaresOption, lock,
+                                        None, change_cls),
                               beginnerContext=(
                                   create, edit, copy, delete, singleRere, rotateLeft, rotateRight, lock, change_cls),
                               advancedContext=(createMode, editMode, edit, copy,
@@ -1105,7 +1107,9 @@ class MainWindow(QMainWindow):
             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:
-            trans_dic = {"label": box[1][0], "points": box[0], "difficult": False, "key_cls": "None"}
+            trans_dic = {"label": box[1][0], "points": box[0], "difficult": False}
+            if self.kie_mode:
+                trans_dic.update({"key_cls": "None"})
             if trans_dic["label"] == "" and mode == 'Auto':
                 continue
             shapes.append(trans_dic)
@@ -1113,8 +1117,10 @@ class MainWindow(QMainWindow):
         try:
             trans_dic = []
             for box in shapes:
-                trans_dic.append({"transcription": box['label'], "points": box['points'],
-                                  "difficult": box['difficult'], "key_cls": box['key_cls']})
+                trans_dict = {"transcription": box['label'], "points": box['points'], "difficult": box['difficult']}
+                if self.kie_mode:
+                    trans_dict.update({"key_cls": box['key_cls']})
+                trans_dic.append(trans_dict)
             self.PPlabel[annotationFilePath] = trans_dic
             if mode == 'Auto':
                 self.Cachelabel[annotationFilePath] = trans_dic
@@ -1424,15 +1430,17 @@ class MainWindow(QMainWindow):
         # 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:
+            key_cls = None if not self.kie_mode else box['key_cls']
             if self.canvas.isInTheSameImage:
                 shapes.append((box['transcription'], [[s[0] * width, s[1] * height] for s in box['ratio']],
-                               DEFAULT_LOCK_COLOR, box['key_cls'], box['difficult']))
+                               DEFAULT_LOCK_COLOR, key_cls, 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, box['key_cls'], box['difficult']))
+                               DEFAULT_LOCK_COLOR, key_cls, box['difficult']))
         if imgidx in self.PPlabel.keys():
             for box in self.PPlabel[imgidx]:
-                shapes.append((box['transcription'], box['points'], None, box['key_cls'], box['difficult']))
+                key_cls = None if not self.kie_mode else box['key_cls']
+                shapes.append((box['transcription'], box['points'], None, key_cls, box['difficult']))
 
         self.loadLabels(shapes)
         self.canvas.verified = False
@@ -1460,6 +1468,7 @@ class MainWindow(QMainWindow):
     def adjustScale(self, initial=False):
         value = self.scalers[self.FIT_WINDOW if initial else self.zoomMode]()
         self.zoomWidget.setValue(int(100 * value))
+        self.imageSlider.setValue(self.zoomWidget.value())  # set zoom slider value
 
     def scaleFitWindow(self):
         """Figure out the size of the pixmap in order to fit the main widget."""
@@ -1600,7 +1609,6 @@ class MainWindow(QMainWindow):
         else:
             self.keyDialog.labelList.addItems(self.existed_key_cls_set)
 
-
     def importDirImages(self, dirpath, isDelete=False):
         if not self.mayContinue() or not dirpath:
             return
@@ -2238,13 +2246,22 @@ class MainWindow(QMainWindow):
             print('The program will automatically save once after confirming 5 images (default)')
 
     def change_box_key(self):
+        if not self.kie_mode:
+            return
         key_text, _ = self.keyDialog.popUp(self.key_previous_text)
         if key_text is None:
             return
         self.key_previous_text = key_text
         for shape in self.canvas.selectedShapes:
             shape.key_cls = key_text
+            if not self.keyList.findItemsByLabel(key_text):
+                item = self.keyList.createItemFromLabel(key_text)
+                self.keyList.addItem(item)
+                rgb = self._get_rgb_by_label(key_text, self.kie_mode)
+                self.keyList.setItemLabel(item, key_text, rgb)
+
             self._update_shape_color(shape)
+            self.keyDialog.addLabelHistory(key_text)
 
     def undoShapeEdit(self):
         self.canvas.restoreShape()
@@ -2288,9 +2305,10 @@ class MainWindow(QMainWindow):
             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'],
-                                  "key_cls": "None" if "key_cls" not in box else box["key_cls"]})
+                trans_dict = {"transcription": box['label'], "ratio": box['ratio'], "difficult": box['difficult']}
+                if self.kie_mode:
+                    trans_dict.update({"key_cls": box["key_cls"]})
+                trans_dic.append(trans_dict)
             self.canvas.lockedShapes = trans_dic
             self.actions.save.setEnabled(True)
 

PPOCRLabel/README.md

@@ -9,7 +9,7 @@ PPOCRLabel is a semi-automatic graphic annotation tool suitable for OCR field, w
 ### Recent Update
 
 - 2022.02：（by [PeterH0323](https://github.com/peterh0323) ）
-  - Added KIE mode, for [detection + identification + keyword extraction] labeling.
+  - Add KIE Mode by using `--kie`, for [detection + identification + keyword extraction] labeling.
 - 2022.01：（by [PeterH0323](https://github.com/peterh0323) ）
   - Improve user experience: prompt for the number of files and labels, optimize interaction, and fix bugs such as only use CPU when inference
 - 2021.11.17:
@@ -54,7 +54,10 @@ PPOCRLabel can be started in two ways: whl package and Python script. The whl pa
 
 ```bash
 pip install PPOCRLabel  # install
-PPOCRLabel  # run
+
+# Select label mode and run 
+PPOCRLabel  # [Normal mode] for [detection + recognition] labeling
+PPOCRLabel --kie True # [KIE mode] for [detection + recognition + keyword extraction] labeling
 ```
 
 > If you getting this error `OSError: [WinError 126] The specified module could not be found` when you install shapely on windows. Please try to download Shapely whl file using http://www.lfd.uci.edu/~gohlke/pythonlibs/#shapely.
@@ -67,13 +70,18 @@ PPOCRLabel  # run
 ```bash
 pip3 install PPOCRLabel
 pip3 install trash-cli
-PPOCRLabel
+
+# Select label mode and run 
+PPOCRLabel  # [Normal mode] for [detection + recognition] labeling
+PPOCRLabel --kie True # [KIE mode] for [detection + recognition + keyword extraction] labeling
 ```
 
 #### MacOS
 ```bash
 pip3 install PPOCRLabel
 pip3 install opencv-contrib-python-headless==4.2.0.32
+
+# Select label mode and run 
 PPOCRLabel  # [Normal mode] for [detection + recognition] labeling
 PPOCRLabel --kie True # [KIE mode] for [detection + recognition + keyword extraction] labeling
 ```
@@ -90,6 +98,8 @@ pip3 install dist/PPOCRLabel-1.0.2-py2.py3-none-any.whl
 
 ```bash
 cd ./PPOCRLabel  # Switch to the PPOCRLabel directory
+
+# Select label mode and run 
 python PPOCRLabel.py  # [Normal mode] for [detection + recognition] labeling
 python PPOCRLabel.py --kie True # [KIE mode] for [detection + recognition + keyword extraction] labeling
 ```
@@ -156,6 +166,7 @@ python PPOCRLabel.py --kie True # [KIE mode] for [detection + recognition + keyw
 | X                        | Rotate the box anti-clockwise                    |
 | C                        | Rotate the box clockwise                         |
 | Ctrl + E                 | Edit label of the selected box                   |
+| Ctrl + X                 | Change key class of the box when enable `--kie`  |
 | Ctrl + R                 | Re-recognize the selected box                    |
 | Ctrl + C                 | Copy and paste the selected box                  |
 | Ctrl + Left Mouse Button | Multi select the label box                       |

PPOCRLabel/README_ch.md

@@ -9,7 +9,7 @@ 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
 
 - 2022.02：（by [PeterH0323](https://github.com/peterh0323) ）
-  - 新增：KIE 功能，用于打【检测+识别+关键字提取】的标签
+  - 新增：使用 `--kie` 进入 KIE 功能，用于打【检测+识别+关键字提取】的标签
 - 2022.01：（by [PeterH0323](https://github.com/peterh0323) ）
   - 提升用户体验：新增文件与标记数目提示、优化交互、修复gpu使用等问题
 - 2021.11.17：
@@ -57,7 +57,10 @@ PPOCRLabel可通过whl包与Python脚本两种方式启动，whl包形式启动
 
 ```bash
 pip install PPOCRLabel  # 安装
-PPOCRLabel --lang ch  # 运行
+
+# 选择标签模式来启动
+PPOCRLabel --lang ch  # 启动【普通模式】，用于打【检测+识别】场景的标签
+PPOCRLabel --lang ch --kie True  # 启动 【KIE 模式】，用于打【检测+识别+关键字提取】场景的标签
 ```
 > 注意：通过whl包安装PPOCRLabel会自动下载 `paddleocr` whl包，其中shapely依赖可能会出现 `[winRrror 126] 找不到指定模块的问题。` 的错误，建议从[这里](https://www.lfd.uci.edu/~gohlke/pythonlibs/#shapely)下载并安装
 ##### Ubuntu Linux
@@ -65,13 +68,18 @@ PPOCRLabel --lang ch  # 运行
 ```bash
 pip3 install PPOCRLabel
 pip3 install trash-cli
-PPOCRLabel --lang ch
+
+# 选择标签模式来启动
+PPOCRLabel --lang ch  # 启动【普通模式】，用于打【检测+识别】场景的标签
+PPOCRLabel --lang ch --kie True  # 启动 【KIE 模式】，用于打【检测+识别+关键字提取】场景的标签
 ```
 
 ##### MacOS
 ```bash
 pip3 install PPOCRLabel
 pip3 install opencv-contrib-python-headless==4.2.0.32 # 如果下载过慢请添加"-i https://mirror.baidu.com/pypi/simple"
+
+# 选择标签模式来启动
 PPOCRLabel --lang ch  # 启动【普通模式】，用于打【检测+识别】场景的标签
 PPOCRLabel --lang ch --kie True  # 启动 【KIE 模式】，用于打【检测+识别+关键字提取】场景的标签
 ```
@@ -92,6 +100,8 @@ pip3 install dist/PPOCRLabel-1.0.2-py2.py3-none-any.whl -i https://mirror.baidu.
 
 ```bash
 cd ./PPOCRLabel  # 切换到PPOCRLabel目录
+
+# 选择标签模式来启动
 python PPOCRLabel.py --lang ch  # 启动【普通模式】，用于打【检测+识别】场景的标签
 python PPOCRLabel.py --lang ch --kie True  # 启动 【KIE 模式】，用于打【检测+识别+关键字提取】场景的标签
 ```
@@ -138,13 +148,14 @@ python PPOCRLabel.py --lang ch --kie True  # 启动 【KIE 模式】，用于打
 ### 3.1 快捷键
 
 | 快捷键              | 说明                              |
-|------------------|----------------|
+|------------------|---------------------------------|
 | 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                           |
 | X                | 框逆时针旋转                          |
 | C                | 框顺时针旋转                          |
 | Ctrl + E         | 编辑所选框标签                         |
+| Ctrl + X         |  `--kie` 模式下，修改 Box 的关键字种类 |
 | 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                           |
@@ -157,6 +168,7 @@ python PPOCRLabel.py --lang ch --kie True  # 启动 【KIE 模式】，用于打
 | Ctrl--           | 放大                              |
 | ↑→↓←             | 移动标记框                           |
 
+
 ### 3.2 内置模型
 
  - 默认模型：PPOCRLabel默认使用PaddleOCR中的中英文超轻量OCR模型，支持中英文与数字识别，多种语言检测。

PPOCRLabel/libs/canvas.py

@@ -546,7 +546,7 @@ class Canvas(QWidget):
         # Give up if both fail.
         for shape in shapes:
             point = shape[0]
-            offset = QPointF(2.0, 2.0)
+            offset = QPointF(5.0, 5.0)
             self.calculateOffsets(shape, point)
             self.prevPoint = point
             if not self.boundedMoveShape(shape, point - offset):

PPOCRLabel/libs/unique_label_qlist_widget.py

 # -*- encoding: utf-8 -*-
 
-from PyQt5.QtCore import Qt
+from PyQt5.QtCore import Qt, QSize
 from PyQt5 import QtWidgets
 
 
@@ -40,6 +40,7 @@ class UniqueLabelQListWidget(EscapableQListWidget):
             qlabel.setText('<font color="#{:02x}{:02x}{:02x}">●</font> {} '.format(*color, label))
         qlabel.setAlignment(Qt.AlignBottom)
 
-        item.setSizeHint(qlabel.sizeHint())
+        # item.setSizeHint(qlabel.sizeHint())
+        item.setSizeHint(QSize(25, 25))
 
         self.setItemWidget(item, qlabel)

README_ch.md

@@ -32,7 +32,7 @@ PaddleOCR旨在打造一套丰富、领先、且实用的OCR工具库，助力
 - PP-OCR系列高质量预训练模型，准确的识别效果
     - 超轻量PP-OCRv2系列：检测（3.1M）+ 方向分类器（1.4M）+ 识别（8.5M）= 13.0M
     - 超轻量PP-OCR mobile移动端系列：检测（3.0M）+方向分类器（1.4M）+ 识别（5.0M）= 9.4M
-    - 通用PPOCR server系列：检测（47.1M）+方向分类器（1.4M）+ 识别（94.9M）= 143.4M
+    - 通用PP-OCR server系列：检测（47.1M）+方向分类器（1.4M）+ 识别（94.9M）= 143.4M
     - 支持中英文数字组合识别、竖排文本识别、长文本识别
     - 支持多语言识别：韩语、日语、德语、法语等约80种语言
 - PP-Structure文档结构化系统

benchmark/run_benchmark_det.sh

 #!/usr/bin/env bash
-set -xe
 # 运行示例：CUDA_VISIBLE_DEVICES=0 bash run_benchmark.sh ${run_mode} ${bs_item} ${fp_item} 500 ${model_mode}
 # 参数说明
 function _set_params(){
@@ -34,11 +33,13 @@ function _train(){
         train_cmd="python tools/train.py "${train_cmd}""
         ;;
       mp)
+        rm -rf ./mylog
         train_cmd="python -m paddle.distributed.launch --log_dir=./mylog --gpus=$CUDA_VISIBLE_DEVICES tools/train.py ${train_cmd}"
         ;;
       *) echo "choose run_mode(sp or mp)"; exit 1;
     esac
 # 以下不用修改
+    echo ${train_cmd}
     timeout 15m ${train_cmd} > ${log_file} 2>&1
     if [ $? -ne 0 ];then
             echo -e "${model_name}, FAIL"

configs/det/det_mv3_pse.yml

@@ -56,7 +56,7 @@ PostProcess:
   thresh: 0
   box_thresh: 0.85
   min_area: 16
-  box_type: box # 'box' or 'poly'
+  box_type: quad # 'quad' or 'poly'
   scale: 1
 
 Metric:

configs/det/det_r50_vd_dcn_fce_ctw.yml

+Global:
+  use_gpu: true
+  epoch_num: 1500
+  log_smooth_window: 20
+  print_batch_step: 20
+  save_model_dir: ./output/det_r50_dcn_fce_ctw/
+  save_epoch_step: 100
+  # evaluation is run every 835 iterations
+  eval_batch_step: [0, 835]
+  cal_metric_during_train: False
+  pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained 
+  checkpoints: 
+  save_inference_dir: 
+  use_visualdl: False
+  infer_img: doc/imgs_en/img_10.jpg
+  save_res_path: ./output/det_fce/predicts_fce.txt
+
+
+Architecture:
+  model_type: det
+  algorithm: FCE
+  Transform:
+  Backbone:
+    name: ResNet
+    layers: 50
+    dcn_stage: [False, True, True, True]
+    out_indices: [1,2,3]
+  Neck:
+    name: FCEFPN
+    out_channels: 256
+    has_extra_convs: False
+    extra_stage: 0
+  Head:
+    name: FCEHead
+    fourier_degree: 5
+Loss:
+  name: FCELoss
+  fourier_degree: 5
+  num_sample: 50
+  
+Optimizer:
+  name: Adam
+  beta1: 0.9
+  beta2: 0.999
+  lr:
+    learning_rate: 0.0001
+  regularizer:
+    name: 'L2'
+    factor: 0
+
+PostProcess:
+  name: FCEPostProcess
+  scales: [8, 16, 32]
+  alpha: 1.0
+  beta: 1.0
+  fourier_degree: 5
+  box_type: 'poly'
+
+Metric:
+  name: DetFCEMetric
+  main_indicator: hmean
+
+Train:
+  dataset:
+    name: SimpleDataSet
+    data_dir: ./train_data/ctw1500/imgs/
+    label_file_list: 
+      - ./train_data/ctw1500/imgs/training.txt
+    transforms:
+      - DecodeImage: # load image
+          img_mode: BGR
+          channel_first: False
+          ignore_orientation: True
+      - DetLabelEncode: # Class handling label
+      - ColorJitter: 
+          brightness: 0.142
+          saturation: 0.5
+          contrast: 0.5
+      - RandomScaling: 
+      - RandomCropFlip:
+          crop_ratio: 0.5
+      - RandomCropPolyInstances:
+          crop_ratio: 0.8
+          min_side_ratio: 0.3
+      - RandomRotatePolyInstances:
+          rotate_ratio: 0.5
+          max_angle: 30
+          pad_with_fixed_color: False
+      - SquareResizePad:
+          target_size: 800
+          pad_ratio: 0.6
+      - IaaAugment:
+          augmenter_args:
+            - { 'type': Fliplr, 'args': { 'p': 0.5 } }
+      - FCENetTargets:
+          fourier_degree: 5
+      - NormalizeImage:
+          scale: 1./255.
+          mean: [0.485, 0.456, 0.406]
+          std: [0.229, 0.224, 0.225]
+          order: 'hwc'
+      - ToCHWImage:
+      - KeepKeys:
+          keep_keys: ['image', 'p3_maps', 'p4_maps', 'p5_maps'] # dataloader will return list in this order
+  loader:
+    shuffle: True
+    drop_last: False
+    batch_size_per_card: 6
+    num_workers: 8
+
+Eval:
+  dataset:
+    name: SimpleDataSet
+    data_dir: ./train_data/ctw1500/imgs/
+    label_file_list:
+      - ./train_data/ctw1500/imgs/test.txt
+    transforms:
+      - DecodeImage: # load image
+          img_mode: BGR
+          channel_first: False
+          ignore_orientation: True
+      - DetLabelEncode: # Class handling label
+      - DetResizeForTest:
+          limit_type: 'min'
+          limit_side_len: 736
+      - NormalizeImage:
+          scale: 1./255.
+          mean: [0.485, 0.456, 0.406]
+          std: [0.229, 0.224, 0.225]
+          order: 'hwc'
+      - Pad: 
+      - ToCHWImage:
+      - KeepKeys:
+          keep_keys: ['image', 'shape', 'polys', 'ignore_tags']
+  loader:
+    shuffle: False
+    drop_last: False
+    batch_size_per_card: 1 # must be 1
+    num_workers: 2
\ No newline at end of file

configs/det/det_r50_vd_pse.yml

@@ -55,7 +55,7 @@ PostProcess:
   thresh: 0
   box_thresh: 0.85
   min_area: 16
-  box_type: box # 'box' or 'poly'
+  box_type: quad # 'quad' or 'poly'
   scale: 1
 
 Metric:

configs/rec/rec_efficientb3_fpn_pren.yml

+Global:
+  use_gpu: True
+  epoch_num: 8
+  log_smooth_window: 20
+  print_batch_step: 5
+  save_model_dir: ./output/rec/pren_new
+  save_epoch_step: 3
+  # evaluation is run every 2000 iterations after the 4000th iteration
+  eval_batch_step: [4000, 2000]
+  cal_metric_during_train: True
+  pretrained_model: 
+  checkpoints:
+  save_inference_dir:
+  use_visualdl: False
+  infer_img: doc/imgs_words/ch/word_1.jpg
+  # for data or label process
+  character_dict_path:
+  max_text_length: &max_text_length 25
+  infer_mode: False
+  use_space_char: False
+  save_res_path: ./output/rec/predicts_pren.txt
+
+Optimizer:
+  name: Adadelta
+  lr:
+    name: Piecewise
+    decay_epochs: [2, 5, 7]
+    values: [0.5, 0.1, 0.01, 0.001]
+
+Architecture:
+  model_type: rec
+  algorithm: PREN
+  in_channels: 3
+  Backbone:
+    name: EfficientNetb3_PREN
+  Neck:
+    name: PRENFPN
+    n_r: 5
+    d_model: 384
+    max_len: *max_text_length
+    dropout: 0.1
+  Head:
+    name: PRENHead
+
+Loss:
+  name: PRENLoss
+
+PostProcess:
+  name: PRENLabelDecode
+
+Metric:
+  name: RecMetric
+  main_indicator: acc
+
+Train:
+  dataset:
+    name: LMDBDataSet
+    data_dir: ./train_data/data_lmdb_release/training/
+    transforms:
+      - DecodeImage:
+          img_mode: BGR
+          channel_first: False
+      - PRENLabelEncode:
+      - RecAug:
+      - PRENResizeImg: 
+          image_shape: [64, 256]  # h,w
+      - KeepKeys:
+          keep_keys: ['image', 'label']
+  loader:
+    shuffle: True
+    batch_size_per_card: 128
+    drop_last: True
+    num_workers: 8
+
+Eval:
+  dataset:
+    name: LMDBDataSet
+    data_dir: ./train_data/data_lmdb_release/validation/
+    transforms:
+      - DecodeImage:
+          img_mode: BGR
+          channel_first: False
+      - PRENLabelEncode:
+      - PRENResizeImg: 
+          image_shape: [64, 256]  # h,w
+      - KeepKeys:
+          keep_keys: ['image', 'label']
+  loader:
+    shuffle: False
+    drop_last: False
+    batch_size_per_card: 64
+    num_workers: 8

deploy/android_demo/README.md

-# 如何快速测试
-### 1. 安装最新版本的Android Studio
-可以从 https://developer.android.com/studio 下载。本Demo使用是4.0版本Android Studio编写。
+- [Android Demo](#android-demo)
+  - [1. 简介](#1-简介)
+  - [2. 近期更新](#2-近期更新)
+  - [3. 快速使用](#3-快速使用)
+    - [3.1 环境准备](#31-环境准备)
+    - [3.2 导入项目](#32-导入项目)
+    - [3.3 运行demo](#33-运行demo)
+    - [3.4 运行模式](#34-运行模式)
+    - [3.5 设置](#35-设置)
+  - [4 更多支持](#4-更多支持)
 
-### 2. 按照NDK 20 以上版本
-Demo测试的时候使用的是NDK 20b版本，20版本以上均可以支持编译成功。
+# Android Demo
 
-如果您是初学者，可以用以下方式安装和测试NDK编译环境。
-点击 File -> New ->New Project，  新建  "Native C++" project
+## 1. 简介
+此为PaddleOCR的Android Demo，目前支持文本检测，文本方向分类器和文本识别模型的使用。使用 [PaddleLite v2.10](https://github.com/PaddlePaddle/Paddle-Lite/tree/release/v2.10) 进行开发。
+
+## 2. 近期更新
+* 2022.02.27
+    * 预测库更新到PaddleLite v2.10
+    * 支持6种运行模式：
+      * 检测+分类+识别
+      * 检测+识别
+      * 分类+识别
+      * 检测
+      * 识别
+      * 分类
+
+## 3. 快速使用
+
+### 3.1 环境准备
+1. 在本地环境安装好 Android Studio 工具，详细安装方法请见[Android Stuido 官网](https://developer.android.com/studio)。
+2. 准备一部 Android 手机，并开启 USB 调试模式。开启方法: `手机设置 -> 查找开发者选项 -> 打开开发者选项和 USB 调试模式`
+
+**注意**：如果您的 Android Studio 尚未配置 NDK ，请根据 Android Studio 用户指南中的[安装及配置 NDK 和 CMake ](https://developer.android.com/studio/projects/install-ndk)内容，预先配置好 NDK 。您可以选择最新的 NDK 版本，或者使用 Paddle Lite 预测库版本一样的 NDK
+
+### 3.2 导入项目
 
-### 3. 导入项目
 点击 File->New->Import Project...， 然后跟着Android Studio的引导导入
+导入完成后呈现如下界面
+![](https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/imgs/import_demo.jpg)
+
+### 3.3 运行demo
+将手机连接上电脑后，点击Android Studio工具栏中的运行按钮即可运行demo。在此过程中，手机会弹出"允许从 USB 安装软件权限"的弹窗，点击允许即可。
+
+软件安转到手机上后会在手机主屏最后一页看到如下app
+<div align="left">
+    <img src="https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/imgs/install_finish.jpeg" width="400">
+</div>
+
+点击app图标即可启动app，启动后app主页如下
+
+<div align="left">
+    <img src="https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/imgs/main_page.jpg" width="400">
+</div>
+
+app主页中有四个按钮，一个下拉列表和一个菜单按钮，他们的功能分别为
+
+* 运行模型：按照已选择的模式，运行对应的模型组合
+* 拍照识别：唤起手机相机拍照并获取拍照的图像，拍照完成后需要点击运行模型进行识别
+* 选取图片：唤起手机相册拍照选择图像，选择完成后需要点击运行模型进行识别
+* 清空绘图：清空当前显示图像上绘制的文本框，以便进行下一次识别(每次识别使用的图像都是当前显示的图像)
+* 下拉列表：进行运行模式的选择，目前包含6种运行模式，默认模式为**检测+分类+识别**详细说明见下一节。
+* 菜单按钮：点击后会进入菜单界面，进行模型和内置图像有关设置
+
+点击运行模型后，会按照所选择的模式运行对应的模型，**检测+分类+识别**模式下运行的模型结果如下所示：
+
+<img src="https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/imgs/run_det_cls_rec.jpg" width="400">
+
+模型运行完成后，模型和运行状态显示区`STATUS`字段显示了当前模型的运行状态，这里显示为`run model successed`表明模型运行成功。
+
+模型的运行结果显示在运行结果显示区，显示格式为
+```text
+序号：Det：(x1,y1)(x2,y2)(x3,y3)(x4,y4) Rec: 识别文本,识别置信度 Cls：分类类别,分类分时
+```
+
+### 3.4 运行模式
+
+PaddleOCR demo共提供了6种运行模式，如下图
+<div align="left">
+    <img src="https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/imgs/select_mode.jpg" width="400">
+</div>
+
+每种模式的运行结果如下表所示
+
+| 检测+分类+识别                                                                                       | 检测+识别                                                                                      | 分类+识别                                                                                      |
+|------------------------------------------------------------------------------------------------|--------------------------------------------------------------------------------------------|--------------------------------------------------------------------------------------------|
+| <img src="https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/imgs/run_det_cls_rec.jpg" width="400"> | <img src="https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/imgs/run_det_rec.jpg" width="400"> | <img src="https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/imgs/run_cls_rec.jpg" width="400"> |
+
+
+| 检测                                                                                     | 识别                                                                                     | 分类                                                                                     |
+|----------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------|
+| <img src="https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/imgs/run_det.jpg" width="400"> | <img src="https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/imgs/run_rec.jpg" width="400"> | <img src="https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/imgs/run_cls.jpg" width="400"> |
+
+### 3.5 设置
+
+设置界面如下
 
+<div align="left">
+    <img src="https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/imgs/settings.jpg" width="400">
+</div>
 
-# 获得更多支持
-前往[端计算模型生成平台EasyEdge](https://ai.baidu.com/easyedge/app/open_source_demo?referrerUrl=paddlelite)，获得更多开发支持：
+在设置界面可以进行如下几项设定：
+1. 普通设置
+   * Enable custom settings: 选中状态下才能更改设置
+   * Model Path: 所运行的模型地址，使用默认值就好
+   * Label Path: 识别模型的字典
+   * Image Path: 进行识别的内置图像名
+2. 模型运行态设置，此项设置更改后返回主界面时，会自动重新加载模型
+   * CPU Thread Num: 模型运行使用的CPU核心数量
+   * CPU Power Mode: 模型运行模式，大小核设定
+3. 输入设置
+   * det long size: DB模型预处理时图像的长边长度，超过此长度resize到该值，短边进行等比例缩放，小于此长度不进行处理。
+4. 输出设置
+   * Score Threshold: DB模型后处理box的阈值，低于此阈值的box进行过滤，不显示。
 
-- Demo APP：可使用手机扫码安装，方便手机端快速体验文字识别
-- SDK：模型被封装为适配不同芯片硬件和操作系统SDK，包括完善的接口，方便进行二次开发
+## 4 更多支持
+1. 实时识别，更新预测库可参考 https://github.com/PaddlePaddle/Paddle-Lite-Demo/tree/develop/ocr/android/app/cxx/ppocr_demo
+2. 更多Paddle-Lite相关问题可前往[Paddle-Lite](https://github.com/PaddlePaddle/Paddle-Lite) ，获得更多开发支持

deploy/android_demo/app/build.gradle

@@ -8,8 +8,8 @@ android {
         applicationId "com.baidu.paddle.lite.demo.ocr"
         minSdkVersion 23
         targetSdkVersion 29
-        versionCode 1
-        versionName "1.0"
+        versionCode 2
+        versionName "2.0"
         testInstrumentationRunner "android.support.test.runner.AndroidJUnitRunner"
         externalNativeBuild {
             cmake {
@@ -17,11 +17,6 @@ android {
                 arguments '-DANDROID_PLATFORM=android-23', '-DANDROID_STL=c++_shared' ,"-DANDROID_ARM_NEON=TRUE"
             }
         }
-        ndk {
-            // abiFilters "arm64-v8a", "armeabi-v7a"
-            abiFilters   "arm64-v8a", "armeabi-v7a"
-            ldLibs "jnigraphics"
-        }
     }
     buildTypes {
         release {
@@ -48,7 +43,7 @@ dependencies {
 
 def archives = [
         [
-                'src' : 'https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/paddle_lite_libs_v2_9_0.tar.gz',
+                'src' : 'https://paddleocr.bj.bcebos.com/libs/paddle_lite_libs_v2_10.tar.gz',
                 'dest': 'PaddleLite'
         ],
         [
@@ -56,7 +51,7 @@ def archives = [
                 'dest': 'OpenCV'
         ],
         [
-                'src' : 'https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ocr_v2_for_cpu.tar.gz',
+                'src' : 'https://paddleocr.bj.bcebos.com/PP-OCRv2/lite/ch_PP-OCRv2.tar.gz',
                 'dest' : 'src/main/assets/models'
         ],
         [

deploy/android_demo/app/src/main/AndroidManifest.xml

@@ -14,7 +14,6 @@
             android:roundIcon="@mipmap/ic_launcher_round"
             android:supportsRtl="true"
             android:theme="@style/AppTheme">
-        <!-- to test MiniActivity, change this to com.baidu.paddle.lite.demo.ocr.MiniActivity -->
         <activity android:name="com.baidu.paddle.lite.demo.ocr.MainActivity">
             <intent-filter>
                 <action android:name="android.intent.action.MAIN"/>