det_mv3_db.yml 3.1 KB
Newer Older
L
LDOUBLEV 已提交
1 2 3 4
Global:
  use_gpu: true
  epoch_num: 1200
  log_smooth_window: 20
W
WenmuZhou 已提交
5
  print_batch_step: 10
W
WenmuZhou 已提交
6
  save_model_dir: ./output/db_mv3/
W
WenmuZhou 已提交
7
  save_epoch_step: 1200
W
update  
WenmuZhou 已提交
8 9
  # evaluation is run every 2000 iterations
  eval_batch_step: [0, 2000]
W
WenmuZhou 已提交
10
  cal_metric_during_train: False
W
WenmuZhou 已提交
11
  pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
W
WenmuZhou 已提交
12
  checkpoints:
L
LDOUBLEV 已提交
13
  save_inference_dir:
D
dyning 已提交
14
  use_visualdl: False
W
WenmuZhou 已提交
15 16
  infer_img: doc/imgs_en/img_10.jpg
  save_res_path: ./output/det_db/predicts_db.txt
L
LDOUBLEV 已提交
17

W
WenmuZhou 已提交
18
Architecture:
D
dyning 已提交
19
  model_type: det
W
WenmuZhou 已提交
20 21 22 23 24 25 26
  algorithm: DB
  Transform:
  Backbone:
    name: MobileNetV3
    scale: 0.5
    model_name: large
  Neck:
D
dyning 已提交
27
    name: DBFPN
W
WenmuZhou 已提交
28 29 30 31
    out_channels: 256
  Head:
    name: DBHead
    k: 50
L
LDOUBLEV 已提交
32 33

Loss:
W
WenmuZhou 已提交
34
  name: DBLoss
L
LDOUBLEV 已提交
35 36 37 38 39
  balance_loss: true
  main_loss_type: DiceLoss
  alpha: 5
  beta: 10
  ohem_ratio: 3
W
WenmuZhou 已提交
40

D
dyning 已提交
41 42 43 44
Optimizer:
  name: Adam
  beta1: 0.9
  beta2: 0.999
W
WenmuZhou 已提交
45 46
  lr:
    learning_rate: 0.001
D
dyning 已提交
47 48 49
  regularizer:
    name: 'L2'
    factor: 0
L
LDOUBLEV 已提交
50 51

PostProcess:
W
WenmuZhou 已提交
52
  name: DBPostProcess
L
LDOUBLEV 已提交
53
  thresh: 0.3
W
WenmuZhou 已提交
54
  box_thresh: 0.6
L
LDOUBLEV 已提交
55
  max_candidates: 1000
W
WenmuZhou 已提交
56 57 58 59 60 61
  unclip_ratio: 1.5

Metric:
  name: DetMetric
  main_indicator: hmean

D
dyning 已提交
62
Train:
W
WenmuZhou 已提交
63 64
  dataset:
    name: SimpleDataSet
D
dyning 已提交
65 66 67
    data_dir: ./train_data/icdar2015/text_localization/
    label_file_list:
      - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt
文幕地方's avatar
文幕地方 已提交
68
    ratio_list: [1.0]
W
WenmuZhou 已提交
69 70 71 72 73 74 75 76
    transforms:
      - DecodeImage: # load image
          img_mode: BGR
          channel_first: False
      - DetLabelEncode: # Class handling label
      - IaaAugment:
          augmenter_args:
            - { 'type': Fliplr, 'args': { 'p': 0.5 } }
D
dyning 已提交
77 78
            - { 'type': Affine, 'args': { 'rotate': [-10, 10] } }
            - { 'type': Resize, 'args': { 'size': [0.5, 3] } }
W
WenmuZhou 已提交
79
      - EastRandomCropData:
D
dyning 已提交
80
          size: [640, 640]
W
WenmuZhou 已提交
81 82 83 84 85 86 87 88 89 90 91
          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.
D
dyning 已提交
92 93
          mean: [0.485, 0.456, 0.406]
          std: [0.229, 0.224, 0.225]
W
WenmuZhou 已提交
94 95
          order: 'hwc'
      - ToCHWImage:
D
dyning 已提交
96 97
      - KeepKeys:
          keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list
W
WenmuZhou 已提交
98
  loader:
W
WenmuZhou 已提交
99
    shuffle: True
W
WenmuZhou 已提交
100
    drop_last: False
W
WenmuZhou 已提交
101
    batch_size_per_card: 16
W
WenmuZhou 已提交
102
    num_workers: 8
103
    use_shared_memory: False
W
WenmuZhou 已提交
104

D
dyning 已提交
105
Eval:
W
WenmuZhou 已提交
106 107
  dataset:
    name: SimpleDataSet
D
dyning 已提交
108 109 110
    data_dir: ./train_data/icdar2015/text_localization/
    label_file_list:
      - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt
W
WenmuZhou 已提交
111 112 113 114 115 116
    transforms:
      - DecodeImage: # load image
          img_mode: BGR
          channel_first: False
      - DetLabelEncode: # Class handling label
      - DetResizeForTest:
D
dyning 已提交
117
          image_shape: [736, 1280]
W
WenmuZhou 已提交
118 119
      - NormalizeImage:
          scale: 1./255.
D
dyning 已提交
120 121
          mean: [0.485, 0.456, 0.406]
          std: [0.229, 0.224, 0.225]
W
WenmuZhou 已提交
122 123
          order: 'hwc'
      - ToCHWImage:
D
dyning 已提交
124 125
      - KeepKeys:
          keep_keys: ['image', 'shape', 'polys', 'ignore_tags']
W
WenmuZhou 已提交
126
  loader:
W
WenmuZhou 已提交
127
    shuffle: False
W
WenmuZhou 已提交
128
    drop_last: False
D
dyning 已提交
129
    batch_size_per_card: 1 # must be 1
130 131
    num_workers: 8
    use_shared_memory: False