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

# Conflicts:
#	PPOCRLabel/libs/resources.py

PPOCRLabel/PPOCRLabel.py

@@ -401,6 +401,7 @@ class MainWindow(QMainWindow, WindowMixin):
         help = action(getStr('tutorial'), self.showTutorialDialog, None, 'help', getStr('tutorialDetail'))
         showInfo = action(getStr('info'), self.showInfoDialog, None, 'help', getStr('info'))
         showSteps = action(getStr('steps'), self.showStepsDialog, None, 'help', getStr('steps'))
+        showKeys = action(getStr('keys'), self.showKeysDialog, None, 'help', getStr('keys'))
 
         zoom = QWidgetAction(self)
         zoom.setDefaultWidget(self.zoomWidget)
@@ -568,7 +569,7 @@ class MainWindow(QMainWindow, WindowMixin):
         addActions(self.menus.file,
                    (opendir, open_dataset_dir, None, saveLabel, saveRec, self.autoSaveOption, None, resetAll, deleteImg, quit))
 
-        addActions(self.menus.help, (showSteps, showInfo))
+        addActions(self.menus.help, (showKeys,showSteps, showInfo))
         addActions(self.menus.view, (
             self.displayLabelOption, self.labelDialogOption,
              None,
@@ -763,6 +764,10 @@ class MainWindow(QMainWindow, WindowMixin):
         msg = stepsInfo(self.lang)
         QMessageBox.information(self, u'Information', msg)
 
+    def showKeysDialog(self):
+        msg = keysInfo(self.lang)
+        QMessageBox.information(self, u'Information', msg)
+
     def createShape(self):
         assert self.beginner()
         self.canvas.setEditing(False)

PPOCRLabel/libs/utils.py

@@ -174,6 +174,7 @@ def stepsInfo(lang='en'):
               "10. 标注结果：关闭应用程序或切换文件路径后，手动保存过的标签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 “PaddleOCR” - 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。\n"
+
     else:
         msg = "1. Build and launch using the instructions above.\n" \
               "2. Click 'Open Dir' in Menu/File to select the folder of the picture.\n"\
@@ -188,4 +189,56 @@ def stepsInfo(lang='en'):
               "9. Click 'Delete Image' and the image will be deleted to the recycle bin.\n"\
               "10. Labeling result: After closing the application or switching the file path, the manually saved label will be stored in *Label.txt* under the opened picture folder.\n"\
               "    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*.\n" 
+
+    return msg
+
+def keysInfo(lang='en'):
+    if lang == 'ch':
+        msg = "快捷键\t\t\t说明\n" \
+              "———————————————————————\n"\
+              "Ctrl + shift + R\t\t对当前图片的所有标记重新识别\n" \
+              "W\t\t\t新建矩形框\n" \
+              "Q\t\t\t新建四点框\n" \
+              "Ctrl + E\t\t编辑所选框标签\n" \
+              "Ctrl + R\t\t重新识别所选标记\n" \
+              "Ctrl + C\t\t复制并粘贴选中的标记框\n" \
+              "Ctrl + 鼠标左键\t\t多选标记框\n" \
+              "Backspace\t\t删除所选框\n" \
+              "Ctrl + V\t\t确认本张图片标记\n" \
+              "Ctrl + Shift + d\t删除本张图片\n" \
+              "D\t\t\t下一张图片\n" \
+              "A\t\t\t上一张图片\n" \
+              "Ctrl++\t\t\t缩小\n" \
+              "Ctrl--\t\t\t放大\n" \
+              "↑→↓←\t\t\t移动标记框\n" \
+              "———————————————————————\n" \
+              "注：Mac用户Command键替换上述Ctrl键"
+
+    else:
+        msg = "Shortcut Keys\t\tDescription\n" \
+              "———————————————————————\n" \
+              "Ctrl + shift + R\t\tRe-recognize all the labels\n" \
+              "\t\t\tof the current image\n" \
+              "\n"\
+              "W\t\t\tCreate a rect box\n" \
+              "Q\t\t\tCreate a four-points box\n" \
+              "Ctrl + E\t\tEdit label of the selected box\n" \
+              "Ctrl + R\t\tRe-recognize the selected box\n" \
+              "Ctrl + C\t\tCopy and paste the selected\n" \
+              "\t\t\tbox\n" \
+              "\n"\
+              "Ctrl + Left Mouse\tMulti select the label\n" \
+              "Button\t\t\tbox\n" \
+              "\n"\
+              "Backspace\t\tDelete the selected box\n" \
+              "Ctrl + V\t\tCheck image\n" \
+              "Ctrl + Shift + d\tDelete image\n" \
+              "D\t\t\tNext image\n" \
+              "A\t\t\tPrevious image\n" \
+              "Ctrl++\t\t\tZoom in\n" \
+              "Ctrl--\t\t\tZoom out\n" \
+              "↑→↓←\t\t\tMove selected box" \
+              "———————————————————————\n" \
+              "Notice:For Mac users, use the 'Command' key instead of the 'Ctrl' key"
+
     return msg
\ No newline at end of file

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

@@ -90,6 +90,7 @@ saveRec=保存识别结果
 tempLabel=待识别
 nullLabel=无法识别
 steps=操作步骤
+keys=快捷键
 choseModelLg=选择模型语言
 cancel=取消
 ok=确认

PPOCRLabel/resources/strings/strings.properties

@@ -90,6 +90,7 @@ saveRec=Save Recognition Result
 tempLabel=TEMPORARY
 nullLabel=NULL
 steps=Steps
+keys=Shortcut Keys
 choseModelLg=Choose Model Language
 cancel=Cancel
 ok=OK

configs/det/ch_ppocr_v2.1/ch_det_lite_train_cml_v2.1.yml

+Global:
+  use_gpu: true
+  epoch_num: 1200
+  log_smooth_window: 20
+  print_batch_step: 2
+  save_model_dir: ./output/ch_db_mv3/
+  save_epoch_step: 1200
+  # evaluation is run every 5000 iterations after the 4000th iteration
+  eval_batch_step: [3000, 2000]
+  cal_metric_during_train: False
+  pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
+  checkpoints:
+  save_inference_dir:
+  use_visualdl: False
+  infer_img: doc/imgs_en/img_10.jpg
+  save_res_path: ./output/det_db/predicts_db.txt
+
+Architecture:
+  name: DistillationModel
+  algorithm: Distillation
+  Models:
+    Student:
+      pretrained: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
+      freeze_params: false
+      return_all_feats: false
+      model_type: det
+      algorithm: DB
+      Backbone:
+        name: MobileNetV3
+        scale: 0.5
+        model_name: large
+        disable_se: True
+      Neck:
+        name: DBFPN
+        out_channels: 96
+      Head:
+        name: DBHead
+        k: 50
+    Student2:
+      pretrained: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
+      freeze_params: false
+      return_all_feats: false
+      model_type: det
+      algorithm: DB
+      Transform:
+      Backbone:
+        name: MobileNetV3
+        scale: 0.5
+        model_name: large
+        disable_se: True
+      Neck:
+        name: DBFPN
+        out_channels: 96
+      Head:
+        name: DBHead
+        k: 50
+    Teacher:
+      pretrained: ./pretrain_models/ch_ppocr_server_v2.0_det_train/best_accuracy
+      freeze_params: true
+      return_all_feats: false
+      model_type: det
+      algorithm: DB
+      Transform:
+      Backbone:
+        name: ResNet
+        layers: 18
+      Neck:
+        name: DBFPN
+        out_channels: 256
+      Head:
+        name: DBHead
+        k: 50
+
+Loss:
+  name: CombinedLoss
+  loss_config_list:
+  - DistillationDilaDBLoss:
+      weight: 1.0
+      model_name_pairs:
+      - ["Student", "Teacher"]
+      - ["Student2", "Teacher"]
+      key: maps
+      balance_loss: true
+      main_loss_type: DiceLoss
+      alpha: 5
+      beta: 10
+      ohem_ratio: 3
+  - DistillationDMLLoss:
+      model_name_pairs:
+      - ["Student", "Student2"]
+      maps_name: "thrink_maps"
+      weight: 1.0
+      # act: None
+      model_name_pairs: ["Student", "Student2"]
+      key: maps
+  - DistillationDBLoss:
+      weight: 1.0
+      model_name_list: ["Student", "Student2"]
+      # key: maps
+      # name: DBLoss
+      balance_loss: true
+      main_loss_type: DiceLoss
+      alpha: 5
+      beta: 10
+      ohem_ratio: 3
+
+
+Optimizer:
+  name: Adam
+  beta1: 0.9
+  beta2: 0.999
+  lr:
+    name: Cosine
+    learning_rate: 0.001
+    warmup_epoch: 2
+  regularizer:
+    name: 'L2'
+    factor: 0
+
+PostProcess:
+  name: DistillationDBPostProcess
+  model_name: ["Student", "Student2", "Teacher"]
+  # key: maps
+  thresh: 0.3
+  box_thresh: 0.6
+  max_candidates: 1000
+  unclip_ratio: 1.5
+
+Metric:
+  name: DistillationMetric
+  base_metric_name: DetMetric
+  main_indicator: hmean
+  key: "Student"
+
+Train:
+  dataset:
+    name: SimpleDataSet
+    data_dir: ./train_data/icdar2015/text_localization/
+    label_file_list:
+      - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt
+    ratio_list: [1.0]
+    transforms:
+      - DecodeImage: # load image
+          img_mode: BGR
+          channel_first: False
+      - DetLabelEncode: # Class handling label
+      - IaaAugment:
+          augmenter_args:
+            - { 'type': Fliplr, 'args': { 'p': 0.5 } }
+            - { 'type': Affine, 'args': { 'rotate': [-10, 10] } }
+            - { 'type': Resize, 'args': { 'size': [0.5, 3] } }
+      - EastRandomCropData:
+          size: [960, 960]
+          max_tries: 50
+          keep_ratio: true
+      - MakeBorderMap:
+          shrink_ratio: 0.4
+          thresh_min: 0.3
+          thresh_max: 0.7
+      - MakeShrinkMap:
+          shrink_ratio: 0.4
+          min_text_size: 8
+      - 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', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list
+  loader:
+    shuffle: True
+    drop_last: False
+    batch_size_per_card: 8
+    num_workers: 4
+
+Eval:
+  dataset:
+    name: SimpleDataSet
+    data_dir: ./train_data/icdar2015/text_localization/
+    label_file_list:
+      - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt
+    transforms:
+      - DecodeImage: # load image
+          img_mode: BGR
+          channel_first: False
+      - DetLabelEncode: # Class handling label
+      - DetResizeForTest:
+#           image_shape: [736, 1280]
+      - 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', 'shape', 'polys', 'ignore_tags']
+  loader:
+    shuffle: False
+    drop_last: False
+    batch_size_per_card: 1 # must be 1
+    num_workers: 2

configs/det/ch_ppocr_v2.1/ch_det_lite_train_distill_v2.1.yml

+Global:
+  use_gpu: true
+  epoch_num: 1200
+  log_smooth_window: 20
+  print_batch_step: 2
+  save_model_dir: ./output/ch_db_mv3/
+  save_epoch_step: 1200
+  # evaluation is run every 5000 iterations after the 4000th iteration
+  eval_batch_step: [3000, 2000]
+  cal_metric_during_train: False
+  pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
+  checkpoints:
+  save_inference_dir:
+  use_visualdl: False
+  infer_img: doc/imgs_en/img_10.jpg
+  save_res_path: ./output/det_db/predicts_db.txt
+
+Architecture:
+  name: DistillationModel
+  algorithm: Distillation
+  Models:
+    Student:
+      pretrained: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
+      freeze_params: false
+      return_all_feats: false
+      model_type: det
+      algorithm: DB
+      Backbone:
+        name: MobileNetV3
+        scale: 0.5
+        model_name: large
+        disable_se: True
+      Neck:
+        name: DBFPN
+        out_channels: 96
+      Head:
+        name: DBHead
+        k: 50
+    Teacher:
+      pretrained: ./pretrain_models/ch_ppocr_server_v2.0_det_train/best_accuracy
+      freeze_params: true
+      return_all_feats: false
+      model_type: det
+      algorithm: DB
+      Transform:
+      Backbone:
+        name: ResNet
+        layers: 18
+      Neck:
+        name: DBFPN
+        out_channels: 256
+      Head:
+        name: DBHead
+        k: 50
+
+Loss:
+  name: CombinedLoss
+  loss_config_list:
+  - DistillationDilaDBLoss:
+      weight: 1.0
+      model_name_pairs:
+      - ["Student", "Teacher"]
+      key: maps
+      balance_loss: true
+      main_loss_type: DiceLoss
+      alpha: 5
+      beta: 10
+      ohem_ratio: 3
+  - DistillationDBLoss:
+      weight: 1.0
+      model_name_list: ["Student", "Teacher"]
+      # key: maps
+      name: DBLoss
+      balance_loss: true
+      main_loss_type: DiceLoss
+      alpha: 5
+      beta: 10
+      ohem_ratio: 3
+
+Optimizer:
+  name: Adam
+  beta1: 0.9
+  beta2: 0.999
+  lr:
+    name: Cosine
+    learning_rate: 0.001
+    warmup_epoch: 2
+  regularizer:
+    name: 'L2'
+    factor: 0
+
+PostProcess:
+  name: DistillationDBPostProcess
+  model_name: ["Student", "Student2"]
+  key: head_out
+  thresh: 0.3
+  box_thresh: 0.6
+  max_candidates: 1000
+  unclip_ratio: 1.5
+
+Metric:
+  name: DistillationMetric
+  base_metric_name: DetMetric
+  main_indicator: hmean
+  key: "Student"
+
+Train:
+  dataset:
+    name: SimpleDataSet
+    data_dir: ./train_data/icdar2015/text_localization/
+    label_file_list:
+      - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt
+    ratio_list: [1.0]
+    transforms:
+      - DecodeImage: # load image
+          img_mode: BGR
+          channel_first: False
+      - DetLabelEncode: # Class handling label
+      - IaaAugment:
+          augmenter_args:
+            - { 'type': Fliplr, 'args': { 'p': 0.5 } }
+            - { 'type': Affine, 'args': { 'rotate': [-10, 10] } }
+            - { 'type': Resize, 'args': { 'size': [0.5, 3] } }
+      - EastRandomCropData:
+          size: [960, 960]
+          max_tries: 50
+          keep_ratio: true
+      - MakeBorderMap:
+          shrink_ratio: 0.4
+          thresh_min: 0.3
+          thresh_max: 0.7
+      - MakeShrinkMap:
+          shrink_ratio: 0.4
+          min_text_size: 8
+      - 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', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list
+  loader:
+    shuffle: True
+    drop_last: False
+    batch_size_per_card: 8
+    num_workers: 4
+
+Eval:
+  dataset:
+    name: SimpleDataSet
+    data_dir: ./train_data/icdar2015/text_localization/
+    label_file_list:
+      - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt
+    transforms:
+      - DecodeImage: # load image
+          img_mode: BGR
+          channel_first: False
+      - DetLabelEncode: # Class handling label
+      - DetResizeForTest:
+#           image_shape: [736, 1280]
+      - 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', 'shape', 'polys', 'ignore_tags']
+  loader:
+    shuffle: False
+    drop_last: False
+    batch_size_per_card: 1 # must be 1
+    num_workers: 2

configs/det/ch_ppocr_v2.1/ch_det_lite_train_dml_v2.1.yml

+Global:
+  use_gpu: true
+  epoch_num: 1200
+  log_smooth_window: 20
+  print_batch_step: 2
+  save_model_dir: ./output/ch_db_mv3/
+  save_epoch_step: 1200
+  # evaluation is run every 5000 iterations after the 4000th iteration
+  eval_batch_step: [3000, 2000]
+  cal_metric_during_train: False
+  pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
+  checkpoints:
+  save_inference_dir:
+  use_visualdl: False
+  infer_img: doc/imgs_en/img_10.jpg
+  save_res_path: ./output/det_db/predicts_db.txt
+
+Architecture:
+  name: DistillationModel
+  algorithm: Distillation
+  Models:
+    Student:
+      pretrained: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
+      freeze_params: false
+      return_all_feats: false
+      model_type: det
+      algorithm: DB
+      Backbone:
+        name: MobileNetV3
+        scale: 0.5
+        model_name: large
+        disable_se: True
+      Neck:
+        name: DBFPN
+        out_channels: 96
+      Head:
+        name: DBHead
+        k: 50
+    Student2:
+      pretrained: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
+      freeze_params: false
+      return_all_feats: false
+      model_type: det
+      algorithm: DB
+      Transform:
+      Backbone:
+        name: MobileNetV3
+        scale: 0.5
+        model_name: large
+        disable_se: True
+      Neck:
+        name: DBFPN
+        out_channels: 96
+      Head:
+        name: DBHead
+        k: 50
+
+
+Loss:
+  name: CombinedLoss
+  loss_config_list:
+  - DistillationDMLLoss:
+      model_name_pairs:
+      - ["Student", "Student2"]
+      maps_name: "thrink_maps"
+      weight: 1.0
+      act: "softmax"
+      model_name_pairs: ["Student", "Student2"]
+      key: maps
+  - DistillationDBLoss:
+      weight: 1.0
+      model_name_list: ["Student", "Student2"]
+      # key: maps
+      name: DBLoss
+      balance_loss: true
+      main_loss_type: DiceLoss
+      alpha: 5
+      beta: 10
+      ohem_ratio: 3
+
+
+Optimizer:
+  name: Adam
+  beta1: 0.9
+  beta2: 0.999
+  lr:
+    name: Cosine
+    learning_rate: 0.001
+    warmup_epoch: 2
+  regularizer:
+    name: 'L2'
+    factor: 0
+
+PostProcess:
+  name: DistillationDBPostProcess
+  model_name: ["Student", "Student2"]
+  key: head_out
+  thresh: 0.3
+  box_thresh: 0.6
+  max_candidates: 1000
+  unclip_ratio: 1.5
+
+Metric:
+  name: DistillationMetric
+  base_metric_name: DetMetric
+  main_indicator: hmean
+  key: "Student"
+
+Train:
+  dataset:
+    name: SimpleDataSet
+    data_dir: ./train_data/icdar2015/text_localization/
+    label_file_list:
+      - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt
+    ratio_list: [1.0]
+    transforms:
+      - DecodeImage: # load image
+          img_mode: BGR
+          channel_first: False
+      - DetLabelEncode: # Class handling label
+      - IaaAugment:
+          augmenter_args:
+            - { 'type': Fliplr, 'args': { 'p': 0.5 } }
+            - { 'type': Affine, 'args': { 'rotate': [-10, 10] } }
+            - { 'type': Resize, 'args': { 'size': [0.5, 3] } }
+      - EastRandomCropData:
+          size: [960, 960]
+          max_tries: 50
+          keep_ratio: true
+      - MakeBorderMap:
+          shrink_ratio: 0.4
+          thresh_min: 0.3
+          thresh_max: 0.7
+      - MakeShrinkMap:
+          shrink_ratio: 0.4
+          min_text_size: 8
+      - 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', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list
+  loader:
+    shuffle: True
+    drop_last: False
+    batch_size_per_card: 8
+    num_workers: 4
+
+Eval:
+  dataset:
+    name: SimpleDataSet
+    data_dir: ./train_data/icdar2015/text_localization/
+    label_file_list:
+      - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt
+    transforms:
+      - DecodeImage: # load image
+          img_mode: BGR
+          channel_first: False
+      - DetLabelEncode: # Class handling label
+      - DetResizeForTest:
+#           image_shape: [736, 1280]
+      - 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', 'shape', 'polys', 'ignore_tags']
+  loader:
+    shuffle: False
+    drop_last: False
+    batch_size_per_card: 1 # must be 1
+    num_workers: 2

deploy/cpp_infer/CMakeLists.txt

@@ -13,7 +13,6 @@ SET(TENSORRT_DIR "" CACHE PATH "Compile demo with TensorRT")
 
 set(DEMO_NAME "ocr_system")
 
-
 macro(safe_set_static_flag)
     foreach(flag_var
         CMAKE_CXX_FLAGS CMAKE_CXX_FLAGS_DEBUG CMAKE_CXX_FLAGS_RELEASE

deploy/cpp_infer/src/clipper.cpp

@@ -668,7 +668,7 @@ void DisposeOutPts(OutPt *&pp) {
 //------------------------------------------------------------------------------
 
 inline void InitEdge(TEdge *e, TEdge *eNext, TEdge *ePrev, const IntPoint &Pt) {
-  std::memset(e, 0, sizeof(TEdge));
+  std::memset(e, int(0), sizeof(TEdge));
   e->Next = eNext;
   e->Prev = ePrev;
   e->Curr = Pt;
@@ -1895,17 +1895,17 @@ void Clipper::InsertLocalMinimaIntoAEL(const cInt botY) {
     TEdge *rb = lm->RightBound;
 
     OutPt *Op1 = 0;
-    if (!lb) {
+    if (!lb || !rb) {
       // nb: don't insert LB into either AEL or SEL
       InsertEdgeIntoAEL(rb, 0);
       SetWindingCount(*rb);
       if (IsContributing(*rb))
         Op1 = AddOutPt(rb, rb->Bot);
-    } else if (!rb) {
-      InsertEdgeIntoAEL(lb, 0);
-      SetWindingCount(*lb);
-      if (IsContributing(*lb))
-        Op1 = AddOutPt(lb, lb->Bot);
+      //} else if (!rb) {
+      //  InsertEdgeIntoAEL(lb, 0);
+      //  SetWindingCount(*lb);
+      //  if (IsContributing(*lb))
+      //    Op1 = AddOutPt(lb, lb->Bot);
       InsertScanbeam(lb->Top.Y);
     } else {
       InsertEdgeIntoAEL(lb, 0);
@@ -2547,13 +2547,13 @@ void Clipper::ProcessHorizontal(TEdge *horzEdge) {
     if (dir == dLeftToRight) {
       maxIt = m_Maxima.begin();
       while (maxIt != m_Maxima.end() && *maxIt <= horzEdge->Bot.X)
-        maxIt++;
+        ++maxIt;
       if (maxIt != m_Maxima.end() && *maxIt >= eLastHorz->Top.X)
         maxIt = m_Maxima.end();
     } else {
       maxRit = m_Maxima.rbegin();
       while (maxRit != m_Maxima.rend() && *maxRit > horzEdge->Bot.X)
-        maxRit++;
+        ++maxRit;
       if (maxRit != m_Maxima.rend() && *maxRit <= eLastHorz->Top.X)
         maxRit = m_Maxima.rend();
     }
@@ -2576,13 +2576,13 @@ void Clipper::ProcessHorizontal(TEdge *horzEdge) {
           while (maxIt != m_Maxima.end() && *maxIt < e->Curr.X) {
             if (horzEdge->OutIdx >= 0 && !IsOpen)
               AddOutPt(horzEdge, IntPoint(*maxIt, horzEdge->Bot.Y));
-            maxIt++;
+            ++maxIt;
           }
         } else {
           while (maxRit != m_Maxima.rend() && *maxRit > e->Curr.X) {
             if (horzEdge->OutIdx >= 0 && !IsOpen)
               AddOutPt(horzEdge, IntPoint(*maxRit, horzEdge->Bot.Y));
-            maxRit++;
+            ++maxRit;
           }
         }
       };

deploy/cpp_infer/src/config.cpp

@@ -21,10 +21,10 @@ std::vector<std::string> OCRConfig::split(const std::string &str,
   std::vector<std::string> res;
   if ("" == str)
     return res;
-  char *strs = new char[str.length() + 1];
+  char strs[str.length() + 1];
   std::strcpy(strs, str.c_str());
 
-  char *d = new char[delim.length() + 1];
+  char d[delim.length() + 1];
   std::strcpy(d, delim.c_str());
 
   char *p = std::strtok(strs, d);

doc/doc_ch/inference.md

@@ -147,12 +147,12 @@ python3 tools/infer/predict_det.py --image_dir="./doc/imgs/00018069.jpg" --det_m
 
 如果输入图片的分辨率比较大，而且想使用更大的分辨率预测，可以设置det_limit_side_len 为想要的值，比如1216：
 ```
-python3 tools/infer/predict_det.py --image_dir="./doc/imgs/2.jpg" --det_model_dir="./inference/det_db/" --det_limit_type=max --det_limit_side_len=1216
+python3 tools/infer/predict_det.py --image_dir="./doc/imgs/1.jpg" --det_model_dir="./inference/det_db/" --det_limit_type=max --det_limit_side_len=1216
 ```
 
 如果想使用CPU进行预测，执行命令如下
 ```
-python3 tools/infer/predict_det.py --image_dir="./doc/imgs/2.jpg" --det_model_dir="./inference/det_db/"  --use_gpu=False
+python3 tools/infer/predict_det.py --image_dir="./doc/imgs/1.jpg" --det_model_dir="./inference/det_db/"  --use_gpu=False
 ```
 
 <a name="DB文本检测模型推理"></a>

doc/doc_en/inference_en.md

@@ -154,12 +154,12 @@ Set as `limit_type='min', det_limit_side_len=960`, it means that the shortest si
 
 If the resolution of the input picture is relatively large and you want to use a larger resolution prediction, you can set det_limit_side_len to the desired value, such as 1216:
 ```
-python3 tools/infer/predict_det.py --image_dir="./doc/imgs/22.jpg" --det_model_dir="./inference/det_db/" --det_limit_type=max --det_limit_side_len=1216
+python3 tools/infer/predict_det.py --image_dir="./doc/imgs/1.jpg" --det_model_dir="./inference/det_db/" --det_limit_type=max --det_limit_side_len=1216
 ```
 
 If you want to use the CPU for prediction, execute the command as follows
 ```
-python3 tools/infer/predict_det.py --image_dir="./doc/imgs/22.jpg" --det_model_dir="./inference/det_db/" --use_gpu=False
+python3 tools/infer/predict_det.py --image_dir="./doc/imgs/1.jpg" --det_model_dir="./inference/det_db/" --use_gpu=False
 ```
 
 <a name="DB_DETECTION"></a>

doc/doc_en/update_en.md

@@ -15,8 +15,6 @@
 - 2020.6.8 Add [datasets](./datasets_en.md) and keep updating
 - 2020.6.5 Support exporting `attention` model to `inference_model`
 - 2020.6.5 Support separate prediction and recognition, output result score
-- 2020.6.5 Support exporting `attention` model to `inference_model`
-- 2020.6.5 Support separate prediction and recognition, output result score
 - 2020.5.30 Provide Lightweight Chinese OCR online experience
 - 2020.5.30 Model prediction and training support on Windows system
 - 2020.5.30 Open source general Chinese OCR model

ppocr/data/simple_dataset.py

@@ -46,6 +46,7 @@ class SimpleDataSet(Dataset):
         self.seed = seed
         logger.info("Initialize indexs of datasets:%s" % label_file_list)
         self.data_lines = self.get_image_info_list(label_file_list, ratio_list)
+        self.check_data()
         self.data_idx_order_list = list(range(len(self.data_lines)))
         if self.mode == "train" and self.do_shuffle:
             self.shuffle_data_random()
@@ -102,16 +103,8 @@ class SimpleDataSet(Dataset):
 
     def __getitem__(self, idx):
         file_idx = self.data_idx_order_list[idx]
-        data_line = self.data_lines[file_idx]
+        data = self.data_lines[file_idx]
         try:
-            data_line = data_line.decode('utf-8')
-            substr = data_line.strip("\n").strip("\r").split(self.delimiter)
-            file_name = substr[0]
-            label = substr[1]
-            img_path = os.path.join(self.data_dir, file_name)
-            data = {'img_path': img_path, 'label': label}
-            if not os.path.exists(img_path):
-                raise Exception("{} does not exist!".format(img_path))
             with open(data['img_path'], 'rb') as f:
                 img = f.read()
                 data['image'] = img
@@ -120,8 +113,8 @@ class SimpleDataSet(Dataset):
         except:
             error_meg = traceback.format_exc()
             self.logger.error(
-                "When parsing line {}, error happened with msg: {}".format(
-                    data_line, error_meg))
+                "When parsing file {} and label {}, error happened with msg: {}".format(
+                    data['img_path'],data['label'], error_meg))
             outs = None
         if outs is None:
             # during evaluation, we should fix the idx to get same results for many times of evaluation.
@@ -132,3 +125,17 @@ class SimpleDataSet(Dataset):
 
     def __len__(self):
         return len(self.data_idx_order_list)
+
+    def check_data(self):
+        new_data_lines = []
+        for data_line in self.data_lines:
+            data_line = data_line.decode('utf-8')
+            substr = data_line.strip("\n").strip("\r").split(self.delimiter)
+            file_name = substr[0]
+            label = substr[1]
+            img_path = os.path.join(self.data_dir, file_name)
+            if os.path.exists(img_path):
+                new_data_lines.append({'img_path': img_path, 'label': label})
+            else:
+                self.logger.info("{} does not exist!".format(img_path))
+        self.data_lines = new_data_lines
\ No newline at end of file

ppocr/losses/basic_loss.py

@@ -54,6 +54,27 @@ class CELoss(nn.Layer):
         return loss
 
 
+class KLJSLoss(object):
+    def __init__(self, mode='kl'):
+        assert mode in ['kl', 'js', 'KL', 'JS'], "mode can only be one of ['kl', 'js', 'KL', 'JS']"
+        self.mode = mode
+
+    def __call__(self, p1, p2, reduction="mean"):
+
+        loss = paddle.multiply(p2, paddle.log( (p2+1e-5)/(p1+1e-5) + 1e-5))
+
+        if self.mode.lower() == "js":
+            loss += paddle.multiply(p1, paddle.log((p1+1e-5)/(p2+1e-5) + 1e-5))
+            loss *= 0.5
+        if reduction == "mean":
+            loss = paddle.mean(loss, axis=[1,2])
+        elif reduction=="none" or reduction is None:
+            return loss 
+        else:
+            loss = paddle.sum(loss, axis=[1,2])
+
+        return loss 
+
 class DMLLoss(nn.Layer):
     """
     DMLLoss
@@ -70,16 +91,20 @@ class DMLLoss(nn.Layer):
         else:
             self.act = None
         
+        self.jskl_loss = KLJSLoss(mode="js")
+
     def forward(self, out1, out2):
         if self.act is not None:
             out1 = self.act(out1)
             out2 = self.act(out2)
-
+        if len(out1.shape) < 2:
             log_out1 = paddle.log(out1)
             log_out2 = paddle.log(out2)
             loss = (F.kl_div(
                 log_out1, out2, reduction='batchmean') + F.kl_div(
                     log_out2, out1, reduction='batchmean')) / 2.0
+        else:
+            loss = self.jskl_loss(out1, out2)
         return loss
 
 

ppocr/losses/combined_loss.py

@@ -17,7 +17,7 @@ import paddle.nn as nn
 
 from .distillation_loss import DistillationCTCLoss
 from .distillation_loss import DistillationDMLLoss
-from .distillation_loss import DistillationDistanceLoss
+from .distillation_loss import DistillationDistanceLoss, DistillationDBLoss, DistillationDilaDBLoss
 
 
 class CombinedLoss(nn.Layer):
@@ -44,15 +44,16 @@ class CombinedLoss(nn.Layer):
 
     def forward(self, input, batch, **kargs):
         loss_dict = {}
+        loss_all = 0.
         for idx, loss_func in enumerate(self.loss_func):
             loss = loss_func(input, batch, **kargs)
             if isinstance(loss, paddle.Tensor):
                 loss = {"loss_{}_{}".format(str(loss), idx): loss}
             weight = self.loss_weight[idx]
-            loss = {
-                "{}_{}".format(key, idx): loss[key] * weight
-                for key in loss
-            }
-            loss_dict.update(loss)
-        loss_dict["loss"] = paddle.add_n(list(loss_dict.values()))
+            for key in loss.keys():
+                if key == "loss":
+                    loss_all += loss[key] * weight
+                else:
+                    loss_dict["{}_{}".format(key, idx)] = loss[key]
+        loss_dict["loss"] = loss_all
         return loss_dict

ppocr/losses/distillation_loss.py

@@ -14,23 +14,76 @@
 
 import paddle
 import paddle.nn as nn
+import numpy as np
+import cv2
 
 from .rec_ctc_loss import CTCLoss
 from .basic_loss import DMLLoss
 from .basic_loss import DistanceLoss
+from .det_db_loss import DBLoss
+from .det_basic_loss import BalanceLoss, MaskL1Loss, DiceLoss
+
+
+def _sum_loss(loss_dict):
+    if "loss" in loss_dict.keys():
+        return loss_dict
+    else:
+        loss_dict["loss"] = 0.
+        for k, value in loss_dict.items():
+            if k == "loss":
+                continue
+            else:
+                loss_dict["loss"] += value
+        return loss_dict
 
 
 class DistillationDMLLoss(DMLLoss):
     """
     """
 
-    def __init__(self, model_name_pairs=[], act=None, key=None,
-                 name="loss_dml"):
+    def __init__(self,
+                 model_name_pairs=[],
+                 act=None,
+                 key=None,
+                 maps_name=None,
+                 name="dml"):
         super().__init__(act=act)
         assert isinstance(model_name_pairs, list)
         self.key = key
-        self.model_name_pairs = model_name_pairs
+        self.model_name_pairs = self._check_model_name_pairs(model_name_pairs)
         self.name = name
+        self.maps_name = self._check_maps_name(maps_name)
+    
+    def _check_model_name_pairs(self, model_name_pairs):
+        if not isinstance(model_name_pairs, list):
+            return []
+        elif isinstance(model_name_pairs[0], list) and isinstance(model_name_pairs[0][0], str):
+            return model_name_pairs
+        else:
+            return [model_name_pairs]
+
+    def _check_maps_name(self, maps_name):
+        if maps_name is None:
+            return None
+        elif type(maps_name) == str:
+            return [maps_name]
+        elif type(maps_name) == list:
+            return [maps_name]
+        else:
+            return None
+
+    def _slice_out(self, outs):
+        new_outs = {}
+        for k in self.maps_name:
+            if k == "thrink_maps":
+                new_outs[k] = outs[:, 0, :, :]
+            elif k == "threshold_maps":
+                new_outs[k] = outs[:, 1, :, :]
+            elif k == "binary_maps":
+                new_outs[k] = outs[:, 2, :, :]
+            else:
+                continue
+        return new_outs
 
     def forward(self, predicts, batch):
         loss_dict = dict()
@@ -40,6 +93,8 @@ class DistillationDMLLoss(DMLLoss):
             if self.key is not None:
                 out1 = out1[self.key]
                 out2 = out2[self.key]
+
+            if self.maps_name is None:
                 loss = super().forward(out1, out2)
                 if isinstance(loss, dict):
                     for key in loss:
@@ -47,6 +102,21 @@ class DistillationDMLLoss(DMLLoss):
                                                        idx)] = loss[key]
                 else:
                     loss_dict["{}_{}".format(self.name, idx)] = loss
+            else:
+                outs1 = self._slice_out(out1)
+                outs2 = self._slice_out(out2)
+                for _c, k in enumerate(outs1.keys()):
+                    loss = super().forward(outs1[k], outs2[k])
+                    if isinstance(loss, dict):
+                        for key in loss:
+                            loss_dict["{}_{}_{}_{}_{}".format(key, pair[
+                                0], pair[1], map_name, idx)] = loss[key]
+                    else:
+                        loss_dict["{}_{}_{}".format(self.name, self.maps_name[_c],
+                                                    idx)] = loss
+        
+        loss_dict = _sum_loss(loss_dict)
+
         return loss_dict
 
 
@@ -73,6 +143,98 @@ class DistillationCTCLoss(CTCLoss):
         return loss_dict
 
 
+class DistillationDBLoss(DBLoss):
+    def __init__(self,
+                 model_name_list=[],
+                 balance_loss=True,
+                 main_loss_type='DiceLoss',
+                 alpha=5,
+                 beta=10,
+                 ohem_ratio=3,
+                 eps=1e-6,
+                 name="db",
+                 **kwargs):
+        super().__init__()
+        self.model_name_list = model_name_list
+        self.name = name
+        self.key = None
+
+    def forward(self, predicts, batch):
+        loss_dict = {}
+        for idx, model_name in enumerate(self.model_name_list):
+            out = predicts[model_name]
+            if self.key is not None:
+                out = out[self.key]
+            loss = super().forward(out, batch)
+
+            if isinstance(loss, dict):
+                for key in loss.keys():
+                    if key == "loss":
+                        continue
+                    name = "{}_{}_{}".format(self.name, model_name, key)
+                    loss_dict[name] = loss[key]
+            else:
+                loss_dict["{}_{}".format(self.name, model_name)] = loss
+
+        loss_dict = _sum_loss(loss_dict)
+        return loss_dict
+
+
+class DistillationDilaDBLoss(DBLoss):
+    def __init__(self,
+                 model_name_pairs=[],
+                 key=None,
+                 balance_loss=True,
+                 main_loss_type='DiceLoss',
+                 alpha=5,
+                 beta=10,
+                 ohem_ratio=3,
+                 eps=1e-6,
+                 name="dila_dbloss"):
+        super().__init__()
+        self.model_name_pairs = model_name_pairs
+        self.name = name
+        self.key = key
+
+    def forward(self, predicts, batch):
+        loss_dict = dict()
+        for idx, pair in enumerate(self.model_name_pairs):
+            stu_outs = predicts[pair[0]]
+            tch_outs = predicts[pair[1]]
+            if self.key is not None:
+                stu_preds = stu_outs[self.key]
+                tch_preds = tch_outs[self.key]
+
+            stu_shrink_maps = stu_preds[:, 0, :, :]
+            stu_binary_maps = stu_preds[:, 2, :, :]
+
+            # dilation to teacher prediction
+            dilation_w = np.array([[1, 1], [1, 1]])
+            th_shrink_maps = tch_preds[:, 0, :, :]
+            th_shrink_maps = th_shrink_maps.numpy() > 0.3  # thresh = 0.3 
+            dilate_maps = np.zeros_like(th_shrink_maps).astype(np.float32)
+            for i in range(th_shrink_maps.shape[0]):
+                dilate_maps[i] = cv2.dilate(
+                    th_shrink_maps[i, :, :].astype(np.uint8), dilation_w)
+            th_shrink_maps = paddle.to_tensor(dilate_maps)
+
+            label_threshold_map, label_threshold_mask, label_shrink_map, label_shrink_mask = batch[
+                1:]
+
+            # calculate the shrink map loss
+            bce_loss = self.alpha * self.bce_loss(
+                stu_shrink_maps, th_shrink_maps, label_shrink_mask)
+            loss_binary_maps = self.dice_loss(stu_binary_maps, th_shrink_maps,
+                                              label_shrink_mask)
+
+            # k = f"{self.name}_{pair[0]}_{pair[1]}"
+            k = "{}_{}_{}".format(self.name, pair[0], pair[1])
+            loss_dict[k] = bce_loss + loss_binary_maps
+
+        loss_dict = _sum_loss(loss_dict)
+        return loss_dict
+
+
 class DistillationDistanceLoss(DistanceLoss):
     """
     """

ppocr/metrics/det_metric.py

@@ -55,6 +55,7 @@ class DetMetric(object):
             result = self.evaluator.evaluate_image(gt_info_list, det_info_list)
             self.results.append(result)
 
+
     def get_metric(self):
         """
         return metrics {