| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109 |
- Global:
- use_gpu: true
- epoch_num: 5
- log_smooth_window: 20
- print_batch_step: 20
- save_model_dir: ./output/rec/rec_r31_robustscanner/
- save_epoch_step: 1
- # evaluation is run every 2000 iterations
- eval_batch_step: [0, 2000]
- cal_metric_during_train: True
- pretrained_model:
- checkpoints:
- save_inference_dir:
- use_visualdl: False
- infer_img: doc/imgs_words_en/word_10.png
- # for data or label process
- character_dict_path: ppocr/utils/dict90.txt
- max_text_length: &max_text_length 40
- infer_mode: False
- use_space_char: False
- rm_symbol: True
- save_res_path: ./output/rec/predicts_robustscanner.txt
- Optimizer:
- name: Adam
- beta1: 0.9
- beta2: 0.999
- lr:
- name: Piecewise
- decay_epochs: [3, 4]
- values: [0.001, 0.0001, 0.00001]
- regularizer:
- name: 'L2'
- factor: 0
- Architecture:
- model_type: rec
- algorithm: RobustScanner
- Transform:
- Backbone:
- name: ResNet31
- init_type: KaimingNormal
- Head:
- name: RobustScannerHead
- enc_outchannles: 128
- hybrid_dec_rnn_layers: 2
- hybrid_dec_dropout: 0
- position_dec_rnn_layers: 2
- start_idx: 91
- mask: True
- padding_idx: 92
- encode_value: False
- max_text_length: *max_text_length
- Loss:
- name: SARLoss
- PostProcess:
- name: SARLabelDecode
- Metric:
- name: RecMetric
- is_filter: True
- Train:
- dataset:
- name: LMDBDataSet
- data_dir: ./train_data/data_lmdb_release/training/
- transforms:
- - DecodeImage: # load image
- img_mode: BGR
- channel_first: False
- - SARLabelEncode: # Class handling label
- - RobustScannerRecResizeImg:
- image_shape: [3, 48, 48, 160] # h:48 w:[48,160]
- width_downsample_ratio: 0.25
- max_text_length: *max_text_length
- - KeepKeys:
- keep_keys: ['image', 'label', 'valid_ratio', 'word_positons'] # dataloader will return list in this order
- loader:
- shuffle: True
- batch_size_per_card: 64
- drop_last: True
- num_workers: 8
- use_shared_memory: False
- Eval:
- dataset:
- name: LMDBDataSet
- data_dir: ./train_data/data_lmdb_release/evaluation/
- transforms:
- - DecodeImage: # load image
- img_mode: BGR
- channel_first: False
- - SARLabelEncode: # Class handling label
- - RobustScannerRecResizeImg:
- image_shape: [3, 48, 48, 160]
- max_text_length: *max_text_length
- width_downsample_ratio: 0.25
- - KeepKeys:
- keep_keys: ['image', 'label', 'valid_ratio', 'word_positons'] # dataloader will return list in this order
- loader:
- shuffle: False
- drop_last: False
- batch_size_per_card: 64
- num_workers: 4
- use_shared_memory: False
-
|
