怎么训练完，效果还不如默认的模型，😓，大佬方便看下有什么问题吗？

[564142183]

请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem

• 系统环境/System Environment：docker，registry.baidubce.com/paddlepaddle/paddle:2.6.1-gpu-cuda12.0-cudnn8.9-trt8.6
• 版本号/Version：Paddle：2.6.1 PaddleOCR：2.7.5 问题相关组件/Related components：
• 运行指令/Command Code：
• 完整报错/Complete Error Message：
• 两次的识别结果
  

稍等附上训练过程

[564142183]

数据集：https://aistudio.baidu.com/datasetdetail/107509
用PPOCRLabel标记了24张图片，如下

1.划分数据集

python PPOCRLabel/gen_ocr_train_val_test.py --trainValTestRatio 6:2:2 --datasetRootPath ./train_data --detRootPath ./train_data/det --recRootPath ./train_data/rec

Label.txt

ocr/0000.jpg [{"transcription": "吴运", "points": [[94, 55], [139, 55], [139, 80], [94, 80]], "difficult": false, "key_cls": "name"}, {"transcription": "性别机器人民族汉", "points": [[42, 84], [218, 84], [218, 111], [42, 111]], "difficult": false, "key_cls": "sex"}, {"transcription": "2006年9月11日", "points": [[88, 116], [245, 116], [245, 140], [88, 140]], "difficult": false, "key_cls": "birth"}, {"transcription": "广东省广州某某666号—白制数据集", "points": [[91, 150], [279, 150], [279, 193], [91, 193]], "difficult": false, "key_cls": "adress"}, {"transcription": "213984748281422633", "points": [[144, 240], [311, 240], [311, 260], [144, 260]], "difficult": false, "key_cls": "card"}] ocr/0001.jpg [{"transcription": "何汊", "points": [[95, 55], [136, 55], [136, 78], [95, 78]], "difficult": false, "key_cls": "name"}, {"transcription": "性别女民族汉", "points": [[47, 88], [205, 88], [205, 108], [47, 108]], "difficult": false, "key_cls": "sex"}, {"transcription": "1998年5月6日", "points": [[91, 118], [237, 118], [237, 139], [91, 139]], "difficult": false, "key_cls": "birth"}, {"transcription": "湖南省长沙某某666号—自制数据集", "points": [[90, 153], [280, 153], [280, 197], [90, 197]], "difficult": false, "key_cls": "adress"}, {"transcription": "081494794611860592", "points": [[145, 235], [310, 235], [310, 260], [145, 260]], "difficult": false, "key_cls": "card"}] ocr/0002.jpg [{"transcription": "雷专鼢", "points": [[94, 54], [157, 54], [157, 81], [94, 81]], "difficult": false, "key_cls": "name"}, {"transcription": "性别机器人民族汉", "points": [[45, 88], [207, 88], [207, 107], [45, 107]], "difficult": false, "key_cls": "sex"}, {"transcription": "2018年11月24日", "points": [[90, 117], [239, 117], [239, 141], [90, 141]], "difficult": false, "key_cls": "birth"}, {"transcription": "安徽省合肥某某666号—自制数据集", "points": [[91, 151], [278, 151], [278, 192], [91, 192]], "difficult": false, "key_cls": "adress"}, {"transcription": "965073295312179882", "points": [[145, 238], [310, 238], [310, 258], [145, 258]], "difficult": false, "key_cls": "card"}] ocr/0003.jpg [{"transcription": "卫肮", "points": [[92, 55], [145, 55], [145, 81], [92, 81]], "difficult": false, "key_cls": "name"}, {"transcription": "性别女民族汉", "points": [[45, 88], [207, 88], [207, 106], [45, 106]], "difficult": false, "key_cls": "sex"}, {"transcription": "2019年12月17日", "points": [[88, 116], [238, 116], [238, 137], [88, 137]], "difficult": false, "key_cls": "birth"}, {"transcription": "河南省郑州某某666号—自制数据集", "points": [[90, 153], [277, 153], [277, 192], [90, 192]], "difficult": false, "key_cls": "adress"}, {"transcription": "163878699991251580", "points": [[144, 238], [314, 238], [314, 258], [144, 258]], "difficult": false, "key_cls": "card"}] ocr/0004.jpg [{"transcription": "郝颁", "points": [[93, 56], [144, 56], [144, 81], [93, 81]], "difficult": false, "key_cls": "name"}, {"transcription": "性别男民族汉", "points": [[46, 88], [207, 88], [207, 107], [46, 107]], "difficult": false, "key_cls": "sex"}, {"transcription": "1984年9月19日", "points": [[90, 115], [237, 115], [237, 139], [90, 139]], "difficult": false, "key_cls": "birth"}, {"transcription": "浙江省杭州某某666号—自制数据集", "points": [[90, 154], [278, 154], [278, 193], [90, 193]], "difficult": false, "key_cls": "adress"}, {"transcription": "803226740643271224", "points": [[144, 236], [315, 236], [315, 262], [144, 262]], "difficult": false, "key_cls": "card"}] ocr/0005.jpg [{"transcription": "郝嫖", "points": [[93, 54], [144, 54], [144, 82], [93, 82]], "difficult": false, "key_cls": "name"}, {"transcription": "性别机器人民族汉", "points": [[46, 86], [208, 86], [208, 111], [46, 111]], "difficult": false, "key_cls": "sex"}, {"transcription": "1979年7月17日", "points": [[90, 118], [239, 118], [239, 137], [90, 137]], "difficult": false, "key_cls": "birth"}, {"transcription": "湖北省武汉某某666号一自制数据集", "points": [[91, 152], [277, 152], [277, 193], [91, 193]], "difficult": false, "key_cls": "adress"}, {"transcription": "974132654656644874", "points": [[144, 237], [311, 237], [311, 260], [144, 260]], "difficult": false, "key_cls": "card"}] ocr/0006.jpg [{"transcription": "茅钇鲅", "points": [[93, 56], [156, 56], [156, 77], [93, 77]], "difficult": false, "key_cls": "name"}, {"transcription": "性别未知民族汉", "points": [[46, 88], [207, 88], [207, 107], [46, 107]], "difficult": false, "key_cls": "sex"}, {"transcription": "1978年2月26日", "points": [[89, 118], [241, 118], [241, 139], [89, 139]], "difficult": false, "key_cls": "birth"}, {"transcription": "福建省福州某某666号—自制数据集", "points": [[89, 152], [279, 152], [279, 196], [89, 196]], "difficult": false, "key_cls": "adress"}, {"transcription": "267124840234334444", "points": [[145, 237], [310, 237], [310, 257], [145, 257]], "difficult": false, "key_cls": "card"}] ocr/0007.jpg [{"transcription": "薛氅沂", "points": [[92, 55], [155, 55], [155, 78], [92, 78]], "difficult": false, "key_cls": "name"}, {"transcription": "性别男民族汉", "points": [[45, 87], [207, 87], [207, 106], [45, 106]], "difficult": false, "key_cls": "sex"}, {"transcription": "1995年7月14日", "points": [[88, 118], [236, 118], [236, 138], [88, 138]], "difficult": false, "key_cls": "birth"}, {"transcription": "黑龙江省哈尔滨某某666号自制数据集", "points": [[90, 151], [280, 151], [280, 196], [90, 196]], "difficult": false, "key_cls": "adress"}, {"transcription": "556010053044157446", "points": [[143, 237], [315, 237], [315, 262], [143, 262]], "difficult": false, "key_cls": "card"}] ocr/0008.jpg [{"transcription": "殷蓊", "points": [[93, 53], [144, 53], [144, 82], [93, 82]], "difficult": false, "key_cls": "name"}, {"transcription": "性别女民族汉", "points": [[46, 88], [207, 88], [207, 107], [46, 107]], "difficult": false, "key_cls": "sex"}, {"transcription": "2017年8月6日", "points": [[89, 116], [237, 116], [237, 139], [89, 139]], "difficult": false, "key_cls": "birth"}, {"transcription": "河北省石家庄某某666号一自制数据集", "points": [[91, 151], [279, 151], [279, 194], [91, 194]], "difficult": false, "key_cls": "adress"}, {"transcription": "20190257831503175X", "points": [[145, 237], [314, 237], [314, 260], [145, 260]], "difficult": false, "key_cls": "card"}] ocr/0009.jpg [{"transcription": "熊杼", "points": [[95, 55], [139, 55], [139, 80], [95, 80]], "difficult": false, "key_cls": "name"}, {"transcription": "性别机器人民族汉", "points": [[45, 88], [215, 88], [215, 109], [45, 109]], "difficult": false, "key_cls": "sex"}, {"transcription": "2013年4月5日", "points": [[89, 116], [237, 116], [237, 139], [89, 139]], "difficult": false, "key_cls": "birth"}, {"transcription": "广东省广州某某666号—自制数据集", "points": [[91, 151], [278, 151], [278, 192], [91, 192]], "difficult": false, "key_cls": "adress"}, {"transcription": "661546442301175555", "points": [[145, 236], [312, 236], [312, 258], [145, 258]], "difficult": false, "key_cls": "card"}] ocr/0010.jpg [{"transcription": "计疆怏", "points": [[97, 56], [160, 56], [160, 80], [97, 80]], "difficult": false, "key_cls": "name"}, {"transcription": "性别未知民族汉", "points": [[46, 87], [207, 87], [207, 107], [46, 107]], "difficult": false, "key_cls": "sex"}, {"transcription": "1985年10月17日", "points": [[89, 117], [240, 117], [240, 142], [89, 142]], "difficult": false, "key_cls": "birth"}, {"transcription": "浙江省杭州某某666号—自制数据集", "points": [[91, 151], [278, 151], [278, 195], [91, 195]], "difficult": false, "key_cls": "adress"}, {"transcription": "449688818089602352", "points": [[145, 235], [313, 235], [313, 261], [145, 261]], "difficult": false, "key_cls": "card"}] ocr/0011.jpg [{"transcription": "萧杀鼬", "points": [[94, 55], [155, 55], [155, 78], [94, 78]], "difficult": false, "key_cls": "name"}, {"transcription": "性别未知民族汉", "points": [[46, 86], [214, 86], [214, 111], [46, 111]], "difficult": false, "key_cls": "sex"}, {"transcription": "1986年9月3日", "points": [[90, 117], [240, 117], [240, 139], [90, 139]], "difficult": false, "key_cls": "birth"}, {"transcription": "福建省福州某某666号——自制数据集", "points": [[93, 149], [278, 149], [278, 192], [93, 192]], "difficult": true, "key_cls": "adress"}, {"transcription": "984445587021926237", "points": [[146, 236], [311, 236], [311, 259], [146, 259]], "difficult": false, "key_cls": "card"}] ocr/0012.jpg [{"transcription": "乐骶脲", "points": [[95, 55], [158, 55], [158, 79], [95, 79]], "difficult": false, "key_cls": "name"}, {"transcription": "性别机器人民族汉", "points": [[43, 84], [207, 84], [207, 107], [43, 107]], "difficult": false, "key_cls": "sex"}, {"transcription": "2014年6月2日", "points": [[90, 117], [228, 117], [228, 135], [90, 135]], "difficult": false, "key_cls": "birth"}, {"transcription": "安徽省合肥某某666号—自制数据集", "points": [[89, 149], [279, 149], [279, 194], [89, 194]], "difficult": false, "key_cls": "adress"}, {"transcription": "294336746086848019", "points": [[145, 237], [310, 237], [310, 260], [145, 260]], "difficult": false, "key_cls": "card"}] ocr/0013.jpg [{"transcription": "金桂羔", "points": [[94, 55], [160, 55], [160, 83], [94, 83]], "difficult": false, "key_cls": "name"}, {"transcription": "性别男民族汉", "points": [[45, 87], [207, 87], [207, 108], [45, 108]], "difficult": false, "key_cls": "sex"}, {"transcription": "2019年6月9日", "points": [[89, 117], [237, 117], [237, 140], [89, 140]], "difficult": false, "key_cls": "birth"}, {"transcription": "福建省福州某某666号——自制数据集", "points": [[90, 151], [279, 151], [279, 194], [90, 194]], "difficult": false, "key_cls": "adress"}, {"transcription": "770140527439237475", "points": [[144, 237], [314, 237], [314, 260], [144, 260]], "difficult": false, "key_cls": "card"}] ocr/0014.jpg [{"transcription": "苗停", "points": [[92, 55], [136, 55], [136, 80], [92, 80]], "difficult": false, "key_cls": "name"}, {"transcription": "性别未知民族汉", "points": [[42, 85], [207, 85], [207, 107], [42, 107]], "difficult": false, "key_cls": "sex"}, {"transcription": "2004年2月25日", "points": [[89, 115], [229, 115], [229, 135], [89, 135]], "difficult": false, "key_cls": "birth"}, {"transcription": "福建省福州某某666号—自制数据集", "points": [[91, 151], [277, 151], [277, 193], [91, 193]], "difficult": false, "key_cls": "adress"}, {"transcription": "352678939043916255", "points": [[144, 238], [311, 238], [311, 262], [144, 262]], "difficult": false, "key_cls": "card"}] ocr/0015.jpg [{"transcription": "安狲", "points": [[93, 54], [141, 54], [141, 78], [93, 78]], "difficult": false, "key_cls": "name"}, {"transcription": "性别机器人民族汉", "points": [[44, 86], [213, 86], [213, 112], [44, 112]], "difficult": false, "key_cls": "sex"}, {"transcription": "1985年9月12日", "points": [[90, 116], [241, 116], [241, 141], [90, 141]], "difficult": false, "key_cls": "birth"}, {"transcription": "山东省济南某某666号——自制数据集", "points": [[90, 151], [279, 151], [279, 193], [90, 193]], "difficult": false, "key_cls": "adress"}, {"transcription": "78953491053071603X", "points": [[145, 238], [312, 238], [312, 261], [145, 261]], "difficult": false, "key_cls": "card"}] ocr/0016.jpg [{"transcription": "成最", "points": [[93, 55], [139, 55], [139, 80], [93, 80]], "difficult": false, "key_cls": "name"}, {"transcription": "性别男民族汉", "points": [[47, 88], [212, 88], [212, 109], [47, 109]], "difficult": false, "key_cls": "sex"}, {"transcription": "2004年10月16日", "points": [[91, 118], [238, 118], [238, 139], [91, 139]], "difficult": false, "key_cls": "birth"}, {"transcription": "潮北省武汉某某666号——自制数据集", "points": [[92, 153], [279, 153], [279, 195], [92, 195]], "difficult": false, "key_cls": "adress"}, {"transcription": "585127964734198547", "points": [[145, 238], [311, 238], [311, 260], [145, 260]], "difficult": false, "key_cls": "card"}] ocr/0017.jpg [{"transcription": "凤靓", "points": [[94, 56], [141, 56], [141, 80], [94, 80]], "difficult": false, "key_cls": "name"}, {"transcription": "性别未知民族汉", "points": [[47, 88], [211, 88], [211, 109], [47, 109]], "difficult": false, "key_cls": "sex"}, {"transcription": "1995年6月27日", "points": [[91, 117], [245, 117], [245, 142], [91, 142]], "difficult": false, "key_cls": "birth"}, {"transcription": "山西省太原某某666号—自制数据集", "points": [[92, 153], [279, 153], [279, 193], [92, 193]], "difficult": false, "key_cls": "adress"}, {"transcription": "925371718911532719", "points": [[145, 238], [311, 238], [311, 260], [145, 260]], "difficult": false, "key_cls": "card"}] ocr/0018.jpg [{"transcription": "张辉", "points": [[93, 55], [139, 55], [139, 79], [93, 79]], "difficult": false, "key_cls": "name"}, {"transcription": "性别未知民族汉", "points": [[45, 87], [214, 87], [214, 109], [45, 109]], "difficult": false, "key_cls": "sex"}, {"transcription": "2009年12月6日", "points": [[88, 117], [237, 117], [237, 140], [88, 140]], "difficult": false, "key_cls": "birth"}, {"transcription": "潮北省武汉某某666号—自制数据集", "points": [[93, 152], [278, 152], [278, 194], [93, 194]], "difficult": false, "key_cls": "adress"}, {"transcription": "898135533101984568", "points": [[145, 238], [310, 238], [310, 258], [145, 258]], "difficult": false, "key_cls": "card"}] ocr/0019.jpg [{"transcription": "吕课", "points": [[95, 57], [137, 57], [137, 79], [95, 79]], "difficult": false, "key_cls": "name"}, {"transcription": "性别未知民族汉", "points": [[46, 87], [210, 87], [210, 110], [46, 110]], "difficult": false, "key_cls": "sex"}, {"transcription": "2014年1月16日", "points": [[91, 116], [240, 116], [240, 140], [91, 140]], "difficult": false, "key_cls": "birth"}, {"transcription": "山东省济南某某666号—自制数据集", "points": [[92, 151], [280, 151], [280, 194], [92, 194]], "difficult": false, "key_cls": "adress"}, {"transcription": "418547080384305138", "points": [[146, 235], [310, 235], [310, 260], [146, 260]], "difficult": false, "key_cls": "card"}] ocr/0020.jpg [{"transcription": "祁峁", "points": [[94, 55], [135, 55], [135, 79], [94, 79]], "difficult": false, "key_cls": "name"}, {"transcription": "性别女民族汉", "points": [[46, 88], [209, 88], [209, 111], [46, 111]], "difficult": false, "key_cls": "sex"}, {"transcription": "2002年3月20日", "points": [[90, 116], [238, 116], [238, 140], [90, 140]], "difficult": false, "key_cls": "birth"}, {"transcription": "河南省郑州某某666号——自制数据集", "points": [[92, 151], [277, 151], [277, 193], [92, 193]], "difficult": false, "key_cls": "adress"}, {"transcription": "081155252519463497", "points": [[144, 237], [312, 237], [312, 260], [144, 260]], "difficult": false, "key_cls": "card"}] ocr/0021.jpg [{"transcription": "魏喜糨", "points": [[96, 55], [154, 55], [154, 77], [96, 77]], "difficult": false, "key_cls": "name"}, {"transcription": "性别男民族汉", "points": [[49, 89], [209, 89], [209, 109], [49, 109]], "difficult": false, "key_cls": "sex"}, {"transcription": "2005年7月15日", "points": [[91, 117], [237, 117], [237, 140], [91, 140]], "difficult": false, "key_cls": "birth"}, {"transcription": "福建省福州某某666号—自制数据集", "points": [[93, 152], [279, 152], [279, 195], [93, 195]], "difficult": false, "key_cls": "adress"}, {"transcription": "885898805473809799", "points": [[145, 238], [310, 238], [310, 257], [145, 257]], "difficult": false, "key_cls": "card"}] ocr/0022.jpg [{"transcription": "425763470037534625", "points": [[146, 239], [310, 239], [310, 258], [146, 258]], "difficult": false, "key_cls": "card"}, {"transcription": "吉林省长春某某666号——自制数据集", "points": [[93, 153], [280, 153], [280, 194], [93, 194]], "difficult": false, "key_cls": "adress"}, {"transcription": "1986年2月13日", "points": [[93, 117], [237, 117], [237, 138], [93, 138]], "difficult": false, "key_cls": "birth"}, {"transcription": "性别男民族汉", "points": [[47, 88], [207, 88], [207, 109], [47, 109]], "difficult": false, "key_cls": "sex"}, {"transcription": "傅嘁馐", "points": [[95, 57], [155, 57], [155, 78], [95, 78]], "difficult": false, "key_cls": "name"}] ocr/0023.jpg [{"transcription": "孔操肆", "points": [[97, 57], [155, 57], [155, 80], [97, 80]], "difficult": false, "key_cls": "name"}, {"transcription": "性别男民族汉", "points": [[45, 85], [207, 85], [207, 109], [45, 109]], "difficult": false, "key_cls": "sex"}, {"transcription": "2006年2月21日", "points": [[90, 116], [232, 116], [232, 136], [90, 136]], "difficult": false, "key_cls": "birth"}, {"transcription": "福建省福州号——自制数据集", "points": [[93, 152], [278, 152], [278, 193], [93, 193]], "difficult": false, "key_cls": "adress"}, {"transcription": "760297342529942985", "points": [[146, 239], [309, 239], [309, 257], [146, 257]], "difficult": false, "key_cls": "card"}]

2.训练det模型

python tools/train.py -c configs/det/ch_PP-OCRv4/ch_PP-OCRv4_det_cml.yml

cal_metric_during_train改成false，Train.loader.batch_size_per_card改成10，Global.pretrained_model改成本地模型

Global:
  debug: false
  use_gpu: true
  epoch_num: 500
  log_smooth_window: 20
  print_batch_step: 20
  save_model_dir: ./output/ch_PP-OCRv4_det
  save_epoch_step: 50
  eval_batch_step:
  - 0
  - 1000
  cal_metric_during_train: false
  checkpoints: null
  pretrained_model: ./train/model/ch_PP-OCRv4_det_train/best_accuracy.pdparams
  save_inference_dir: null
  use_visualdl: false
  infer_img: doc/imgs_en/img_10.jpg
  save_res_path: ./checkpoints/det_db/predicts_db.txt
  distributed: true
Architecture:
  name: DistillationModel
  algorithm: Distillation
  model_type: det
  Models:
    Student:
      model_type: det
      algorithm: DB
      Transform: null
      Backbone:
        name: PPLCNetV3
        scale: 0.75
        pretrained: false
        det: true
      Neck:
        name: RSEFPN
        out_channels: 96
        shortcut: true
      Head:
        name: DBHead
        k: 50
    Student2:
      pretrained: null
      model_type: det
      algorithm: DB
      Transform: null
      Backbone:
        name: PPLCNetV3
        scale: 0.75
        pretrained: true
        det: true
      Neck:
        name: RSEFPN
        out_channels: 96
        shortcut: true
      Head:
        name: DBHead
        k: 50
    Teacher:
      pretrained: https://paddleocr.bj.bcebos.com/PP-OCRv4/chinese/ch_PP-OCRv4_det_cml_teacher_pretrained/teacher.pdparams
      freeze_params: true
      return_all_feats: false
      model_type: det
      algorithm: DB
      Backbone:
        name: ResNet_vd
        in_channels: 3
        layers: 50
      Neck:
        name: LKPAN
        out_channels: 256
      Head:
        name: DBHead
        kernel_list:
        - 7
        - 2
        - 2
        k: 50
Loss:
  name: CombinedLoss
  loss_config_list:
  - DistillationDilaDBLoss:
      weight: 1.0
      model_name_pairs:
      - - Student
        - Teacher
      - - Student2
        - Teacher
      key: maps
      balance_loss: true
      main_loss_type: DiceLoss
      alpha: 5
      beta: 10
      ohem_ratio: 3
  - DistillationDMLLoss:
      model_name_pairs:
      - Student
      - Student2
      maps_name: thrink_maps
      weight: 1.0
      key: maps
  - DistillationDBLoss:
      weight: 1.0
      model_name_list:
      - Student
      - Student2
      balance_loss: true
      main_loss_type: DiceLoss
      alpha: 5
      beta: 10
      ohem_ratio: 3
Optimizer:
  name: Adam
  beta1: 0.9
  beta2: 0.999
  lr:
    name: Cosine
    learning_rate: 0.001
    warmup_epoch: 2
  regularizer:
    name: L2
    factor: 5.0e-05
PostProcess:
  name: DistillationDBPostProcess
  model_name:
  - Student
  key: head_out
  thresh: 0.3
  box_thresh: 0.6
  max_candidates: 1000
  unclip_ratio: 1.5
Metric:
  name: DistillationMetric
  base_metric_name: DetMetric
  main_indicator: hmean
  key: Student
Train:
  dataset:
    name: SimpleDataSet
    data_dir: ./train_data/
    label_file_list:
      - ./train_data/det/train.txt
    ratio_list: [1.0]
    transforms:
    - DecodeImage:
        img_mode: BGR
        channel_first: false
    - DetLabelEncode: null
    - IaaAugment:
        augmenter_args:
        - type: Fliplr
          args:
            p: 0.5
        - type: Affine
          args:
            rotate:
            - -10
            - 10
        - type: Resize
          args:
            size:
            - 0.5
            - 3
    - EastRandomCropData:
        size:
        - 640
        - 640
        max_tries: 50
        keep_ratio: true
    - MakeBorderMap:
        shrink_ratio: 0.4
        thresh_min: 0.3
        thresh_max: 0.7
        total_epoch: 500
    - MakeShrinkMap:
        shrink_ratio: 0.4
        min_text_size: 8
        total_epoch: 500
    - NormalizeImage:
        scale: 1./255.
        mean:
        - 0.485
        - 0.456
        - 0.406
        std:
        - 0.229
        - 0.224
        - 0.225
        order: hwc
    - ToCHWImage: null
    - KeepKeys:
        keep_keys:
        - image
        - threshold_map
        - threshold_mask
        - shrink_map
        - shrink_mask
  loader:
    shuffle: true
    drop_last: false
    # batch_size_per_card: 16
    batch_size_per_card: 10
    num_workers: 8
Eval:
  dataset:
    name: SimpleDataSet
    data_dir: ./train_data/
    label_file_list:
      - ./train_data/det/val.txt
    transforms:
    - DecodeImage:
        img_mode: BGR
        channel_first: false
    - DetLabelEncode: null
    - DetResizeForTest: 
        limit_side_len: 960
        limit_type: max
    - NormalizeImage:
        scale: 1./255.
        mean:
        - 0.485
        - 0.456
        - 0.406
        std:
        - 0.229
        - 0.224
        - 0.225
        order: hwc
    - ToCHWImage: null
    - KeepKeys:
        keep_keys:
        - image
        - shape
        - polys
        - ignore_tags
  loader:
    shuffle: false
    drop_last: false
    batch_size_per_card: 1
    num_workers: 2
profiler_options: null

3.训练rec模型

python tools/train.py -c configs/rec/PP-OCRv4/ch_PP-OCRv4_rec.yml

Global.pretrained_model改成本地模型

Global:
  debug: false
  use_gpu: true
  epoch_num: 200
  log_smooth_window: 20
  print_batch_step: 10
  save_model_dir: ./output/ch_PP-OCRv4_rec
  save_epoch_step: 10
  eval_batch_step: [0, 2000]
  cal_metric_during_train: true
  pretrained_model: ./train/model/ch_PP-OCRv4_rec_train/student.pdparams
  checkpoints:
  save_inference_dir:
  use_visualdl: false
  infer_img: doc/imgs_words/ch/word_1.jpg
  character_dict_path: ppocr/utils/ppocr_keys_v1.txt
  max_text_length: &max_text_length 25
  infer_mode: false
  use_space_char: true
  distributed: true
  save_res_path: ./output/rec/predicts_ppocrv3.txt


Optimizer:
  name: Adam
  beta1: 0.9
  beta2: 0.999
  lr:
    name: Cosine
    learning_rate: 0.001
    warmup_epoch: 5
  regularizer:
    name: L2
    factor: 3.0e-05


Architecture:
  model_type: rec
  algorithm: SVTR_LCNet
  Transform:
  Backbone:
    name: PPLCNetV3
    scale: 0.95
  Head:
    name: MultiHead
    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: *max_text_length

Loss:
  name: MultiLoss
  loss_config_list:
    - CTCLoss:
    - NRTRLoss:

PostProcess:  
  name: CTCLabelDecode

Metric:
  name: RecMetric
  main_indicator: acc

Train:
  dataset:
    name: MultiScaleDataSet
    ds_width: false
    data_dir: ./train_data/
    ext_op_transform_idx: 1
    label_file_list:
    - ./train_data/rec/train.txt
    transforms:
    - DecodeImage:
        img_mode: BGR
        channel_first: false
    - RecConAug:
        prob: 0.5
        ext_data_num: 2
        image_shape: [48, 320, 3]
        max_text_length: *max_text_length
    - RecAug:
    - MultiLabelEncode:
        gtc_encode: NRTRLabelEncode
    - KeepKeys:
        keep_keys:
        - image
        - label_ctc
        - label_gtc
        - length
        - valid_ratio
  sampler:
    name: MultiScaleSampler
    scales: [[320, 32], [320, 48], [320, 64]]
    first_bs: &bs 192
    fix_bs: false
    divided_factor: [8, 16] # w, h
    is_training: True
  loader:
    shuffle: true
    batch_size_per_card: *bs
    drop_last: true
    num_workers: 8
Eval:
  dataset:
    name: SimpleDataSet
    data_dir: ./train_data
    label_file_list:
    - ./train_data/rec/val.txt
    transforms:
    - DecodeImage:
        img_mode: BGR
        channel_first: false
    - MultiLabelEncode:
        gtc_encode: NRTRLabelEncode
    - RecResizeImg:
        image_shape: [3, 48, 320]
    - KeepKeys:
        keep_keys:
        - image
        - label_ctc
        - label_gtc
        - length
        - valid_ratio
  loader:
    shuffle: false
    drop_last: false
    batch_size_per_card: 128
    num_workers: 4

4.导出推理模型

python tools/export_model.py -c configs/det/ch_PP-OCRv4/ch_PP-OCRv4_det_cml.yml -o Global.pretrained_model=output/ch_PP-OCRv4_det/best_model/model.pdparams Global.save_inference_dir=inference_model/det/

python tools/export_model.py -c configs/rec/PP-OCRv4/ch_PP-OCRv4_rec.yml -o Global.pretrained_model="./output/ch_PP-OCRv4_rec/latest.pdparams" Global.save_inference_dir="./inference_model/rec/"

5.执行脚本

import cv2
  
from paddleocr import PaddleOCR

paddleocr = PaddleOCR(lang='ch', show_log=False, enable_mkldnn=True,
                      det_model_dir='./inference_model/det/Teacher',
                      rec_model_dir='./inference_model/rec'
)
img = cv2.imread('./train_data/0000.jpg')
result = paddleocr.ocr(img)
for i in range(len(result[0])):
    print(result[0][i][1][0])

[564142183]

代码已上传到aistudio，https://aistudio.baidu.com/projectdetail/7900653?sUid=3018256&shared=1&ts=1716016870508

[UserWangZz]

数据量太少了