OpenOCR/configs/rec/visionlan/resnet45_trans_visionlan_LF_1.yml

Global:
  device: gpu
  epoch_num: 10
  log_smooth_window: 20
  print_batch_step: 10
  output_dir: ./output/rec/u14m_filter/resnet45_trans_visionlan_LF1/
  eval_epoch_step: [0, 1]
  eval_batch_step: [0, 500]
  cal_metric_during_train: True
  pretrained_model:
  checkpoints:
  use_tensorboard: false
  infer_img:
  # for data or label process
  character_dict_path: &character_dict_path ./tools/utils/EN_symbol_dict.txt # 96en
  # ./tools/utils/ppocr_keys_v1.txt  # ch
  max_text_length: &max_text_length 25
  use_space_char: &use_space_char False
  save_res_path: ./output/rec/u14m_filter/predicts_resnet45_trans_visionlan_LF1.txt
  grad_clip_val: 20
  use_amp: True

Optimizer:
  name: Adam
  lr: 0.0002 # for 4gpus bs128/gpu
  weight_decay: 0.0

LRScheduler:
  name: MultiStepLR
  milestones: [6]

Architecture:
  model_type: rec
  algorithm: VisionLAN
  Transform:
  Encoder:
    name: ResNet45
    in_channels: 3
    strides: [2, 2, 2, 1, 1]
  Decoder:
    name: VisionLANDecoder
    training_step: &training_step 'LF_1'
    n_position: 256

Loss:
  name: VisionLANLoss
  training_step: *training_step

PostProcess:
  name: VisionLANLabelDecode
  character_dict_path: *character_dict_path
  use_space_char: *use_space_char

Metric:
  name: RecMetric
  main_indicator: acc
  is_filter: True

Train:
  dataset:
    name: LMDBDataSet
    data_dir: ../Union14M-L-LMDB-Filtered
    transforms:
      - DecodeImagePIL: # load image
          img_mode: RGB
      - PARSeqAugPIL:
      - VisionLANLabelEncode:
          character_dict_path: *character_dict_path
          use_space_char: *use_space_char
          max_text_length: *max_text_length
      - RecTVResize:
          image_shape: [64, 256]
          padding: False
      - KeepKeys:
          keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order
  loader:
    shuffle: True
    batch_size_per_card: 128
    drop_last: True
    num_workers: 4

Eval:
  dataset:
    name: LMDBDataSet
    data_dir: ../evaluation
    transforms:
      - DecodeImagePIL: # load image
          img_mode: RGB
      - VisionLANLabelEncode:
          character_dict_path: *character_dict_path
          use_space_char: *use_space_char
          max_text_length: *max_text_length
      - RecTVResize:
          image_shape: [64, 256]
          padding: False
      - KeepKeys:
          keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order
  loader:
    shuffle: False
    drop_last: False
    batch_size_per_card: 128
    num_workers: 2