configs/rec/visionlan/svtrv2_visionlan_LA.yml

Global:
  device: gpu
  epoch_num: 20
  log_smooth_window: 20
  print_batch_step: 10
  output_dir: ./output/rec/u14m_filter/svtrv2_visionlan_LA/
  eval_epoch_step: [0, 1]
  eval_batch_step: [0, 500]
  cal_metric_during_train: True
  pretrained_model:
  # ./output/rec/u14m_filter/svtrv2_visionlan_LF2/best.pth
  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_svtrv2_visionlan_LA.txt
  grad_clip_val: 20
  use_amp: True

Optimizer:
  name: AdamW
  lr: 0.00065 # for 4gpus bs256/gpu
  weight_decay: 0.05
  filter_bias_and_bn: True

LRScheduler:
  name: OneCycleLR
  warmup_epoch: 1.5 # pct_start 0.075*20 = 1.5ep
  cycle_momentum: False

Architecture:
  model_type: rec
  algorithm: VisionLAN
  Transform:
  Encoder:
    name: SVTRv2LNConvTwo33
    use_pos_embed: False
    dims: [128, 256, 384]
    depths: [6, 6, 6]
    num_heads: [4, 8, 12]
    mixer: [['Conv','Conv','Conv','Conv','Conv','Conv'],['Conv','Conv','FGlobal','Global','Global','Global'],['Global','Global','Global','Global','Global','Global']]
    local_k: [[5, 5], [5, 5], [-1, -1]]
    sub_k: [[1, 1], [2, 1], [-1, -1]]
    last_stage: false
    feat2d: True
  Decoder:
    name: VisionLANDecoder
    training_step: &training_step 'LA'
    n_position: 128

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: [32, 128]
          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: 256
    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: [32, 128]
          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: 256
    num_workers: 2