refactor pre- and post-processing modules

configs/det/db_r50_icdar15.yaml

@@ -17,7 +17,7 @@ model:
     name: DBFPN
     out_channels: 256
     bias: False
-    use_asf: False # enable it for DB++ 
+    use_asf: False        # enable it for DB++
   head:
     name: DBHead
     k: 50
@@ -26,11 +26,11 @@ model:
 
 postprocess:
   name: DBPostprocess
-  region_type: 'quad'
-  thresh: 0.3
-  box_thresh: 0.55 # TODO: this value is 0.55 in modelzoo and but 0.7 in paddle
+  output_polygon: False   # whether to output a polygon or a box
+  binary_thresh: 0.3      # binarization threshold
+  box_thresh: 0.6         # box score threshold
   max_candidates: 1000
-  unclip_ratio: 1.5
+  expand_ratio: 1.5       # coefficient for expanding predictions
 
 metric:
   name: DetMetric
@@ -70,12 +70,12 @@ train:
     shuffle: True
     transform_pipeline:
       - DecodeImage:
-          img_mode: BGR
+          img_mode: RGB
           to_float32: False
       - DetLabelEncode:
       - MZResizeByGrid:
           divisor: 32
-          transform_polys: True # originally in modelzoo, it doesn't transform polys
+          transform_polys: True
       - MZRandomScaleByShortSide:
           short_side: 736
       - IaaAugment:
@@ -86,10 +86,10 @@ train:
           max_tries: 100
           min_crop_side_ratio: 0.1
           crop_size: [ 640, 640 ]
-      - MZMakeSegDetectionData:
+      - ShrinkBinaryMap:
           min_text_size: 8
           shrink_ratio: 0.4
-      - MZMakeBorderMap:
+      - BorderMap:
           shrink_ratio: 0.4
           thresh_min: 0.3
           thresh_max: 0.7
@@ -98,13 +98,13 @@ train:
           saturation: 0.5
           to_numpy: True
       - NormalizeImage:
-          bgr_to_rgb: True
+          bgr_to_rgb: False
           is_hwc: True
           mean: imagenet
           std: imagenet
       - ToCHWImage:
     #  the order of the dataloader list, matching the network input and the input labels for the loss function, and optional data for debug/visaulize 
-    output_keys: ['image', 'shrink_map', 'shrink_mask', 'threshold_map', 'threshold_mask'] #'img_path'] 
+    output_keys: ['image', 'binary_map', 'mask', 'thresh_map', 'thresh_mask'] #'img_path']
     #output_keys: ['image'] # for debug op performance
     num_keys_to_net: 1 # num inputs for network forward func in output_keys
 #    keys_for_loss: 4 # num labels for loss func
@@ -113,7 +113,7 @@ train:
     shuffle: True # TODO: tbc
     batch_size: 20
     drop_remainder: False
-    max_rowsize: 20
+    max_rowsize: 32
     num_workers: 10 # TODO: may lead to OOM
 
 eval:
@@ -125,17 +125,20 @@ eval:
     label_file: ic15/text_localization/test/test_icdar2015_label.txt
     #data_dir: /Users/Samit/Data/datasets/ic15/det/test
     #label_file: /Users/Samit/Data/datasets/ic15/det/test/test_icdar2015_label.txt
-    sample_ratio 1.0 ]
+    sample_ratio: [ 1.0 ]
     shuffle: False
     transform_pipeline:
       - DecodeImage:
-          img_mode: BGR
+          img_mode: RGB
           to_float32: False
       - DetLabelEncode:
+      - MZResizeByGrid:
+          divisor: 32
+          transform_polys: True
       - MZScalePad:
           eval_size: [ 736, 1280 ] # h, w
       - NormalizeImage:
-          bgr_to_rgb: True
+          bgr_to_rgb: False
           is_hwc: True
           mean: imagenet
           std: imagenet