DB 检测模型：训练好的模型直接infer效果很好，export为推理模型后，漏检了很多。【急】

[Smile0524]

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：
• 版本号/Version：Paddle： PaddleOCR： 问题相关组件/Related components：
• 运行指令/Command Code：
• 完整报错/Complete Error Message：

请尽量不要包含图片在问题中/Please try to not include the image in the issue.

各位大佬帮忙看看：

同一张图片，训练好的模型直接infer效果很好，export为推理模型后，漏检了很多。参照issues中大家的做法，把代码中DetResizeForTest注释掉，还是没有用。 export时，参数如下：

Global: use_gpu: false epoch_num: 300 log_smooth_window: 20 print_batch_step: 1 save_model_dir: ../../out_model/ch_db_res18/ save_epoch_step: 10

evaluation is run every 5000 iterations after the 4000th iteration

eval_batch_step: [0, 200] cal_metric_during_train: false pretrained_model: checkpoints: save_inference_dir: use_visualdl: False infer_img: doc/imgs_en/img_10.jpg save_res_path: ../../out_model/ch_db_res18/predicts_db.txt

Architecture: model_type: det algorithm: DB Transform: Backbone: name: ResNet_vd layers: 18 disable_se: True Neck: name: DBFPN out_channels: 256 Head: name: DBHead k: 50

Loss: name: DBLoss 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: 0

PostProcess: name: DBPostProcess thresh: 0.3 box_thresh: 0.6 max_candidates: 1000 unclip_ratio: 1.5

Metric: name: DetMetric main_indicator: hmean

Train: dataset: name: SimpleDataSet data_dir: ../../data/seal_detect_ori/det/train label_file_list:

• ../../data/seal_detect_ori/det/train.txt ratio_list: [1.0] transforms:
• DecodeImage: # load image img_mode: BGR channel_first: False
• DetLabelEncode: # Class handling label
• IaaAugment: augmenter_args:
  • { 'type': Fliplr, 'args': { 'p': 0.5 } }
  • { 'type': Affine, 'args': { 'rotate': [-10, 10] } }
  • { 'type': Resize, 'args': { 'size': [0.5, 3] } }
• EastRandomCropData: size: [960, 960] max_tries: 50 keep_ratio: true
• MakeBorderMap: shrink_ratio: 0.4 thresh_min: 0.3 thresh_max: 0.7
• MakeShrinkMap: shrink_ratio: 0.4 min_text_size: 8
• NormalizeImage: scale: 1./255. mean: [0.485, 0.456, 0.406] std: [0.229, 0.224, 0.225] order: 'hwc'
• ToCHWImage:
• KeepKeys: keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list loader: shuffle: True drop_last: False batch_size_per_card: 8 num_workers: 4

Eval: dataset: name: SimpleDataSet data_dir: ../../data/seal_detect_ori/det/val label_file_list:

• ../../data/seal_detect_ori/det/val.txt transforms:
• DecodeImage: # load image img_mode: BGR channel_first: False
• DetLabelEncode: # Class handling label
• DetResizeForTest:

  limit_side_len: 960

  limit_type: max

  image_shape: [736, 1280]

• NormalizeImage: scale: 1./255. mean: [0.485, 0.456, 0.406] std: [0.229, 0.224, 0.225] order: 'hwc'
• ToCHWImage:
• KeepKeys: keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] loader: shuffle: False drop_last: False batch_size_per_card: 1 # must be 1 num_workers: 2

[Smile0524]

已解决，当检测和识别模型同时使用时，在所有检测模型检测到的检测框中，会依据识别结果的分数再去做一次过滤（参数：drop_score）。所以导致表象上的漏检。

[Nanh1860]

已解决，当检测和识别模型同时使用时，在所有检测模型检测到的检测框中，会依据识别结果的分数再去做一次过滤（参数：drop_score）。所以导致表象上的漏检。

can you tell me how to solve that? my data form is page with many data per row, i want to detect and recognize each large row but unfortunately it only can recognize only 2 row top, sometimes 3