MinerU/mineru/model/utils/pytorchocr/utils/resources/arch_config.yaml

ch_ptocr_mobile_v2.0_cls_infer:
  model_type: cls
  algorithm: CLS
  Transform:
  Backbone:
    name: MobileNetV3
    scale: 0.35
    model_name: small
  Neck:
  Head:
    name: ClsHead
    class_dim: 2

Multilingual_PP-OCRv3_det_infer:
  model_type: det
  algorithm: DB
  Transform:
  Backbone:
    name: MobileNetV3
    scale: 0.5
    model_name: large
    disable_se: True
  Neck:
    name: RSEFPN
    out_channels: 96
    shortcut: True
  Head:
    name: DBHead
    k: 50

en_PP-OCRv3_det_infer:
  model_type: det
  algorithm: DB
  Transform:
  Backbone:
    name: MobileNetV3
    scale: 0.5
    model_name: large
    disable_se: True
  Neck:
    name: RSEFPN
    out_channels: 96
    shortcut: True
  Head:
    name: DBHead
    k: 50

ch_PP-OCRv3_det_infer:
  model_type: det
  algorithm: DB
  Transform:
  Backbone:
    name: MobileNetV3
    scale: 0.5
    model_name: large
    disable_se: True
  Neck:
    name: RSEFPN
    out_channels: 96
    shortcut: True
  Head:
    name: DBHead
    k: 50

en_PP-OCRv4_rec_infer:
  model_type: rec
  algorithm: SVTR_LCNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 97 #'blank' + ...(62) + ' '
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

ch_PP-OCRv4_det_infer:
  model_type: det
  algorithm: DB
  Transform: null
  Backbone:
    name: PPLCNetV3
    scale: 0.75
    det: True
  Neck:
    name: RSEFPN
    out_channels: 96
    shortcut: True
  Head:
    name: DBHead
    k: 50

ch_PP-OCRv5_det_infer:
  model_type: det
  algorithm: DB
  Transform: null
  Backbone:
    name: PPLCNetV3
    scale: 0.75
    det: True
  Neck:
    name: RSEFPN
    out_channels: 96
    shortcut: True
  Head:
    name: DBHead
    k: 50

ch_PP-OCRv5_det_server_infer:
    model_type: det
    algorithm: DB
    Transform: null
    Backbone:
        name: PPHGNetV2_B4
        det: True
    Neck:
        name: LKPAN
        out_channels: 256
        intracl: True
    Head:
        name: PFHeadLocal
        k: 50
        mode: "large"

ch_PP-OCRv4_det_server_infer:
  model_type: det
  algorithm: DB
  Transform: null
  Backbone:
    name: PPHGNet_small
    det: True
  Neck:
    name: LKPAN
    out_channels: 256
    intracl: true
  Head:
    name: PFHeadLocal
    k: 50
    mode: "large"

ch_PP-OCRv4_rec_infer:
  model_type: rec
  algorithm: SVTR_LCNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 6625 #'blank' + ...(6623) + ' '
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

ch_PP-OCRv4_rec_server_infer:
  model_type: rec
  algorithm: SVTR_HGNet
  Transform:
  Backbone:
    name: PPHGNet_small
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 6625 #'blank' + ...(6623) + ' '
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

ch_PP-OCRv4_rec_server_doc_infer:
  model_type: rec
  algorithm: SVTR_HGNet
  Transform:
  Backbone:
    name: PPHGNet_small
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 15631
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

ch_PP-OCRv5_rec_server_infer:
  model_type: rec
  algorithm: SVTR_HGNet
  Transform:
  Backbone:
    name: PPHGNetV2_B4
    text_rec: True
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 18385
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

ch_PP-OCRv5_rec_infer:
  model_type: rec
  algorithm: SVTR_HGNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 18385
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

chinese_cht_PP-OCRv3_rec_infer:
  model_type: rec
  algorithm: SVTR
  Transform:
  Backbone:
    name: MobileNetV1Enhance
    scale: 0.5
    last_conv_stride: [1, 2]
    last_pool_type: avg
  Neck:
    name: SequenceEncoder
    encoder_type: svtr
    dims: 64
    depth: 2
    hidden_dims: 120
    use_guide: True
  Head:
    name: CTCHead
#    out_channels: 8423
    fc_decay: 0.00001

latin_PP-OCRv3_rec_infer:
  model_type: rec
  algorithm: SVTR
  Transform:
  Backbone:
    name: MobileNetV1Enhance
    scale: 0.5
    last_conv_stride: [ 1, 2 ]
    last_pool_type: avg
  Neck:
    name: SequenceEncoder
    encoder_type: svtr
    dims: 64
    depth: 2
    hidden_dims: 120
    use_guide: True
  Head:
    name: CTCHead
#    out_channels: 187
    fc_decay: 0.00001

cyrillic_PP-OCRv3_rec_infer:
  model_type: rec
  algorithm: SVTR
  Transform:
  Backbone:
    name: MobileNetV1Enhance
    scale: 0.5
    last_conv_stride: [ 1, 2 ]
    last_pool_type: avg
  Neck:
    name: SequenceEncoder
    encoder_type: svtr
    dims: 64
    depth: 2
    hidden_dims: 120
    use_guide: True
  Head:
    name: CTCHead
#    out_channels: 165
    fc_decay: 0.00001

arabic_PP-OCRv3_rec_infer:
  model_type: rec
  algorithm: SVTR
  Transform:
  Backbone:
    name: MobileNetV1Enhance
    scale: 0.5
    last_conv_stride: [ 1, 2 ]
    last_pool_type: avg
  Neck:
    name: SequenceEncoder
    encoder_type: svtr
    dims: 64
    depth: 2
    hidden_dims: 120
    use_guide: True
  Head:
    name: CTCHead
#    out_channels: 164
    fc_decay: 0.00001

korean_PP-OCRv3_rec_infer:
  model_type: rec
  algorithm: SVTR
  Transform:
  Backbone:
    name: MobileNetV1Enhance
    scale: 0.5
    last_conv_stride: [ 1, 2 ]
    last_pool_type: avg
  Neck:
    name: SequenceEncoder
    encoder_type: svtr
    dims: 64
    depth: 2
    hidden_dims: 120
    use_guide: True
  Head:
    name: CTCHead
#    out_channels: 3690
    fc_decay: 0.00001

japan_PP-OCRv3_rec_infer:
  model_type: rec
  algorithm: SVTR
  Transform:
  Backbone:
    name: MobileNetV1Enhance
    scale: 0.5
    last_conv_stride: [ 1, 2 ]
    last_pool_type: avg
  Neck:
    name: SequenceEncoder
    encoder_type: svtr
    dims: 64
    depth: 2
    hidden_dims: 120
    use_guide: True
  Head:
    name: CTCHead
#    out_channels: 4401
    fc_decay: 0.00001

ta_PP-OCRv3_rec_infer:
  model_type: rec
  algorithm: SVTR
  Transform:
  Backbone:
    name: MobileNetV1Enhance
    scale: 0.5
    last_conv_stride: [ 1, 2 ]
    last_pool_type: avg
  Neck:
    name: SequenceEncoder
    encoder_type: svtr
    dims: 64
    depth: 2
    hidden_dims: 120
    use_guide: True
  Head:
    name: CTCHead
#    out_channels: 130
    fc_decay: 0.00001

te_PP-OCRv3_rec_infer:
  model_type: rec
  algorithm: SVTR
  Transform:
  Backbone:
    name: MobileNetV1Enhance
    scale: 0.5
    last_conv_stride: [ 1, 2 ]
    last_pool_type: avg
  Neck:
    name: SequenceEncoder
    encoder_type: svtr
    dims: 64
    depth: 2
    hidden_dims: 120
    use_guide: True
  Head:
    name: CTCHead
#    out_channels: 153
    fc_decay: 0.00001

ka_PP-OCRv3_rec_infer:
  model_type: rec
  algorithm: SVTR
  Transform:
  Backbone:
    name: MobileNetV1Enhance
    scale: 0.5
    last_conv_stride: [ 1, 2 ]
    last_pool_type: avg
  Neck:
    name: SequenceEncoder
    encoder_type: svtr
    dims: 64
    depth: 2
    hidden_dims: 120
    use_guide: True
  Head:
    name: CTCHead
#    out_channels: 155
    fc_decay: 0.00001

devanagari_PP-OCRv3_rec_infer:
  model_type: rec
  algorithm: SVTR
  Transform:
  Backbone:
    name: MobileNetV1Enhance
    scale: 0.5
    last_conv_stride: [ 1, 2 ]
    last_pool_type: avg
  Neck:
    name: SequenceEncoder
    encoder_type: svtr
    dims: 64
    depth: 2
    hidden_dims: 120
    use_guide: True
  Head:
    name: CTCHead
#    out_channels: 169
    fc_decay: 0.00001

korean_PP-OCRv5_rec_infer:
  model_type: rec
  algorithm: SVTR_HGNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 11947
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

latin_PP-OCRv5_rec_infer:
  model_type: rec
  algorithm: SVTR_HGNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 504
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

eslav_PP-OCRv5_rec_infer:
  model_type: rec
  algorithm: SVTR_HGNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 519
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

el_PP-OCRv5_rec_infer:
  model_type: rec
  algorithm: SVTR_LCNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 356
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

th_PP-OCRv5_rec_infer:
  model_type: rec
  algorithm: SVTR_LCNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 526
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

en_PP-OCRv5_rec_infer:
  model_type: rec
  algorithm: SVTR_LCNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 438
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

arabic_PP-OCRv5_rec_infer:
  model_type: rec
  algorithm: SVTR_LCNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 749
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

cyrillic_PP-OCRv5_rec_infer:
  model_type: rec
  algorithm: SVTR_LCNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 852
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

devanagari_PP-OCRv5_rec_infer:
   model_type: rec
   algorithm: SVTR_LCNet
   Transform:
   Backbone:
     name: PPLCNetV3
     scale: 0.95
   Head:
     name: MultiHead
     out_channels_list:
       CTCLabelDecode: 570
     head_list:
       - CTCHead:
           Neck:
             name: svtr
             dims: 120
             depth: 2
             hidden_dims: 120
             kernel_size: [ 1, 3 ]
             use_guide: True
           Head:
             fc_decay: 0.00001
       - NRTRHead:
           nrtr_dim: 384
           max_text_length: 25

ta_PP-OCRv5_rec_infer:
  model_type: rec
  algorithm: SVTR_LCNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 515
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25

te_PP-OCRv5_rec_infer:
  model_type: rec
  algorithm: SVTR_LCNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    out_channels_list:
      CTCLabelDecode: 542
    head_list:
      - CTCHead:
          Neck:
            name: svtr
            dims: 120
            depth: 2
            hidden_dims: 120
            kernel_size: [ 1, 3 ]
            use_guide: True
          Head:
            fc_decay: 0.00001
      - NRTRHead:
          nrtr_dim: 384
          max_text_length: 25