Global:
  use_gpu: True
  epoch_num: 400
  log_smooth_window: 20
  print_batch_step: 10
  save_model_dir: ./output/rec/seed
  save_epoch_step: 3
  # evaluation is run every 5000 iterations after the 4000th iteration
  eval_batch_step: [0, 2000]
  cal_metric_during_train: True
  pretrained_model:
  checkpoints:
  save_inference_dir:
  use_visualdl: False
  infer_img: doc/imgs_words_en/word_10.png
  # for data or label process
  character_dict_path: 
  character_type: EN_symbol
  max_text_length: 100
  infer_mode: False
  use_space_char: False
  eval_filter: True
  save_res_path: ./output/rec/predicts_seed.txt


Optimizer:
  name: Adadelta
  weight_deacy: 0.0
  momentum: 0.9
  lr:
    name: Piecewise
    decay_epochs: [4,5,8]
    values: [1.0, 0.1, 0.01]
  regularizer:
    name: 'L2'
    factor: 2.0e-05


Architecture:
  model_type: seed
  algorithm: ASTER
  Transform:
    name: STN_ON
    tps_inputsize: [32, 64]
    tps_outputsize: [32, 100]
    num_control_points: 20
    tps_margins: [0.05,0.05]
    stn_activation: none
  Backbone:
    name: ResNet_ASTER
  Head:
    name: AsterHead  # AttentionHead
    sDim: 512
    attDim: 512
    max_len_labels: 100

Loss:
  name: AsterLoss

PostProcess:
  name: SEEDLabelDecode

Metric:
  name: RecMetric
  main_indicator: acc
  is_filter: True

Train:
  dataset:
    name: LMDBDataSet
    data_dir: ./train_data/data_lmdb_release/training/
    transforms:
      - Fasttext:
          path: "./cc.en.300.bin"
      - DecodeImage: # load image
          img_mode: BGR
          channel_first: False
      - SEEDLabelEncode: # Class handling label
      - SEEDResize:
          image_shape: [3, 64, 256]
      - KeepKeys:
          keep_keys: ['image', 'label', 'length', 'fast_label'] # dataloader will return list in this order
  loader:
    shuffle: True
    batch_size_per_card: 256
    drop_last: True
    num_workers: 6

Eval:
  dataset:
    name: LMDBDataSet
    data_dir: ./train_data/data_lmdb_release/evaluation/
    transforms:
      - DecodeImage: # load image
          img_mode: BGR
          channel_first: False
      - SEEDLabelEncode: # Class handling label
      - SEEDResize:
          image_shape: [3, 64, 256]
      - KeepKeys:
          keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order
  loader:
    shuffle: False
    drop_last: True
    batch_size_per_card: 256
    num_workers: 4