[Config] Update NRTR configs (#1302)

* [Config] Add textrec_default_runtime

* add vis hook

* update nrtr configs

* Update configs/textrecog/nrtr/nrtr_resnet31-1by16-1by8_6e_st_mj.py

Co-authored-by: gaotongxiao <gaotongxiao@gmail.com>

configs/textrecog/nrtr/_base_nrtr_modality-transform.py

@@ -0,0 +1,62 @@
+file_client_args = dict(backend='disk')
+
+dictionary = dict(
+    type='Dictionary',
+    dict_file='dicts/english_digits_symbols.txt',
+    with_padding=True,
+    with_unknown=True,
+    same_start_end=True,
+    with_start=True,
+    with_end=True)
+
+model = dict(
+    type='NRTR',
+    backbone=dict(type='NRTRModalityTransform'),
+    encoder=dict(type='NRTREncoder', n_layers=12),
+    decoder=dict(
+        type='NRTRDecoder',
+        module_loss=dict(
+            type='CEModuleLoss', ignore_first_char=True, flatten=True),
+        postprocessor=dict(type='AttentionPostprocessor'),
+        dictionary=dictionary,
+        max_seq_len=30),
+    data_preprocessor=dict(
+        type='TextRecogDataPreprocessor',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375]))
+
+train_pipeline = [
+    dict(
+        type='LoadImageFromFile',
+        file_client_args=file_client_args,
+        ignore_empty=True,
+        min_size=5),
+    dict(type='LoadOCRAnnotations', with_text=True),
+    dict(
+        type='RescaleToHeight',
+        height=32,
+        min_width=32,
+        max_width=160,
+        width_divisor=4),
+    dict(type='PadToWidth', width=160),
+    dict(
+        type='PackTextRecogInputs',
+        meta_keys=('img_path', 'ori_shape', 'img_shape', 'valid_ratio'))
+]
+
+test_pipeline = [
+    dict(type='LoadImageFromFile', file_client_args=file_client_args),
+    dict(
+        type='RescaleToHeight',
+        height=32,
+        min_width=32,
+        max_width=160,
+        width_divisor=16),
+    dict(type='PadToWidth', width=160),
+    # add loading annotation after ``Resize`` because ground truth
+    # does not need to do resize data transform
+    dict(type='LoadOCRAnnotations', with_text=True),
+    dict(
+        type='PackTextRecogInputs',
+        meta_keys=('img_path', 'ori_shape', 'img_shape', 'valid_ratio'))
+]

configs/textrecog/nrtr/_base_nrtr_resnet31.py

@@ -0,0 +1,68 @@
+file_client_args = dict(backend='disk')
+
+dictionary = dict(
+    type='Dictionary',
+    dict_file='dicts/english_digits_symbols.txt',
+    with_padding=True,
+    with_unknown=True,
+    same_start_end=True,
+    with_start=True,
+    with_end=True)
+
+model = dict(
+    type='NRTR',
+    backbone=dict(
+        type='ResNet31OCR',
+        layers=[1, 2, 5, 3],
+        channels=[32, 64, 128, 256, 512, 512],
+        stage4_pool_cfg=dict(kernel_size=(2, 1), stride=(2, 1)),
+        last_stage_pool=True),
+    encoder=dict(type='NRTREncoder'),
+    decoder=dict(
+        type='NRTRDecoder',
+        module_loss=dict(
+            type='CEModuleLoss', ignore_first_char=True, flatten=True),
+        postprocessor=dict(type='AttentionPostprocessor'),
+        dictionary=dictionary,
+        max_seq_len=30,
+    ),
+    data_preprocessor=dict(
+        type='TextRecogDataPreprocessor',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375]))
+
+train_pipeline = [
+    dict(
+        type='LoadImageFromFile',
+        file_client_args=file_client_args,
+        ignore_empty=True,
+        min_size=5),
+    dict(type='LoadOCRAnnotations', with_text=True),
+    dict(
+        type='RescaleToHeight',
+        height=32,
+        min_width=32,
+        max_width=160,
+        width_divisor=4),
+    dict(type='PadToWidth', width=160),
+    dict(
+        type='PackTextRecogInputs',
+        meta_keys=('img_path', 'ori_shape', 'img_shape', 'valid_ratio'))
+]
+
+test_pipeline = [
+    dict(type='LoadImageFromFile', file_client_args=file_client_args),
+    dict(
+        type='RescaleToHeight',
+        height=32,
+        min_width=32,
+        max_width=160,
+        width_divisor=16),
+    dict(type='PadToWidth', width=160),
+    # add loading annotation after ``Resize`` because ground truth
+    # does not need to do resize data transform
+    dict(type='LoadOCRAnnotations', with_text=True),
+    dict(
+        type='PackTextRecogInputs',
+        meta_keys=('img_path', 'ori_shape', 'img_shape', 'valid_ratio'))
+]

configs/textrecog/nrtr/metafile.yml

@@ -17,9 +17,9 @@ Collections:
   README: configs/textrecog/nrtr/README.md
 
 Models:
-  - Name: nrtr_r31_1by16_1by8_academic
+  - Name: nrtr_resnet31-1by16-1by8_6e_st_mj
     In Collection: NRTR
-    Config: configs/textrecog/nrtr/nrtr_r31_1by16_1by8_academic.py
+    Config: configs/textrecog/nrtr/nrtr_resnet31-1by16-1by8_6e_st_mj.py
     Metadata:
       Training Data:
         - SynthText
@@ -51,9 +51,9 @@ Models:
           word_acc: 87.15
     Weights: https://download.openmmlab.com/mmocr/textrecog/nrtr/nrtr_r31_1by16_1by8_academic_20211124-f60cebf4.pth
 
-  - Name: nrtr_r31_1by8_1by4_academic
+  - Name: nrtr_resnet31-1by8-1by4_6e_st_mj
     In Collection: NRTR
-    Config: configs/textrecog/nrtr/nrtr_r31_1by8_1by4_academic.py
+    Config: configs/textrecog/nrtr/nrtr_resnet31-1by8-1by4_6e_st_mj.py
     Metadata:
       Training Data:
         - SynthText

configs/textrecog/nrtr/nrtr_modality-transform_6e_st_mj.py

@@ -0,0 +1,45 @@
+_base_ = [
+    '../../_base_/recog_datasets/mjsynth.py',
+    '../../_base_/recog_datasets/synthtext.py',
+    '../../_base_/recog_datasets/cute80.py',
+    '../../_base_/recog_datasets/iiit5k.py',
+    '../../_base_/recog_datasets/svt.py',
+    '../../_base_/recog_datasets/svtp.py',
+    '../../_base_/recog_datasets/icdar2013.py',
+    '../../_base_/recog_datasets/icdar2015.py',
+    '../../_base_/textrec_default_runtime.py',
+    '../../_base_/schedules/schedule_adam_step_6e.py',
+    '_base_nrtr_modality-transform.py',
+]
+
+# optimizer settings
+optim_wrapper = dict(type='OptimWrapper', optimizer=dict(type='Adam', lr=3e-4))
+
+# dataset settings
+train_list = [_base_.mj_rec_train, _base_.st_rec_train]
+test_list = [
+    _base_.cute80_rec_test, _base_.iiit5k_rec_test, _base_.svt_rec_test,
+    _base_.svtp_rec_test, _base_.ic13_rec_test, _base_.ic15_rec_test
+]
+
+train_dataset = dict(
+    type='ConcatDataset', datasets=train_list, pipeline=_base_.train_pipeline)
+test_dataset = dict(
+    type='ConcatDataset', datasets=test_list, pipeline=_base_.test_pipeline)
+
+train_dataloader = dict(
+    batch_size=384,
+    num_workers=32,
+    persistent_workers=True,
+    sampler=dict(type='DefaultSampler', shuffle=True),
+    dataset=train_dataset)
+
+test_dataloader = dict(
+    batch_size=1,
+    num_workers=4,
+    persistent_workers=True,
+    drop_last=False,
+    sampler=dict(type='DefaultSampler', shuffle=False),
+    dataset=test_dataset)
+
+val_dataloader = test_dataloader

configs/textrecog/nrtr/nrtr_modality-transform_6e_toy.py

@@ -0,0 +1,41 @@
+_base_ = [
+    '../../_base_/recog_datasets/toy_data.py',
+    '../../_base_/textrec_default_runtime.py',
+    '../../_base_/schedules/schedule_adam_step_6e.py',
+    '_base_nrtr_modality-transform.py',
+]
+
+# dataset settings
+train_list = {{_base_.train_list}}
+test_list = {{_base_.test_list}}
+
+train_dataset = dict(
+    type='ConcatDataset', datasets=train_list, pipeline=_base_.train_pipeline)
+test_dataset = dict(
+    type='ConcatDataset', datasets=test_list, pipeline=_base_.test_pipeline)
+
+train_dataloader = dict(
+    batch_size=8,
+    num_workers=4,
+    persistent_workers=True,
+    sampler=dict(type='DefaultSampler', shuffle=True),
+    dataset=train_dataset)
+
+val_dataloader = dict(
+    batch_size=1,
+    num_workers=4,
+    persistent_workers=True,
+    drop_last=False,
+    sampler=dict(type='DefaultSampler', shuffle=False),
+    dataset=test_dataset)
+
+test_dataloader = val_dataloader
+
+val_evaluator = [
+    dict(
+        type='WordMetric', mode=['exact', 'ignore_case',
+                                 'ignore_case_symbol']),
+    dict(type='CharMetric')
+]
+
+test_evaluator = val_evaluator

configs/textrecog/nrtr/nrtr_modality_transform.py

@@ -1,24 +0,0 @@
-dictionary = dict(
-    type='Dictionary',
-    dict_file='dicts/english_digits_symbols.txt',
-    with_padding=True,
-    with_unknown=True,
-    same_start_end=True,
-    with_start=True,
-    with_end=True)
-
-model = dict(
-    type='NRTR',
-    backbone=dict(type='NRTRModalityTransform'),
-    encoder=dict(type='NRTREncoder', n_layers=12),
-    decoder=dict(
-        type='NRTRDecoder',
-        module_loss=dict(
-            type='CEModuleLoss', ignore_first_char=True, flatten=True),
-        postprocessor=dict(type='AttentionPostprocessor'),
-        dictionary=dictionary,
-        max_seq_len=30),
-    data_preprocessor=dict(
-        type='TextRecogDataPreprocessor',
-        mean=[123.675, 116.28, 103.53],
-        std=[58.395, 57.12, 57.375]))

configs/textrecog/nrtr/nrtr_modality_transform_academic.py

@@ -1,93 +0,0 @@
-_base_ = [
-    '../../_base_/recog_datasets/mjsynth.py',
-    '../../_base_/recog_datasets/synthtext.py',
-    '../../_base_/recog_datasets/cute80.py',
-    '../../_base_/recog_datasets/iiit5k.py',
-    '../../_base_/recog_datasets/svt.py',
-    '../../_base_/recog_datasets/svtp.py',
-    '../../_base_/recog_datasets/icdar2013.py',
-    '../../_base_/recog_datasets/icdar2015.py',
-    '../../_base_/default_runtime.py',
-    '../../_base_/schedules/schedule_adam_step_6e.py',
-    'nrtr_modality_transform.py',
-]
-
-# optimizer settings
-optim_wrapper = dict(type='OptimWrapper', optimizer=dict(type='Adam', lr=3e-4))
-
-# dataset settings
-train_list = [_base_.mj_rec_train, _base_.st_rec_train]
-test_list = [
-    _base_.cute80_rec_test, _base_.iiit5k_rec_test, _base_.svt_rec_test,
-    _base_.svtp_rec_test, _base_.ic13_rec_test, _base_.ic15_rec_test
-]
-
-file_client_args = dict(backend='disk')
-default_hooks = dict(logger=dict(type='LoggerHook', interval=50), )
-
-train_pipeline = [
-    dict(
-        type='LoadImageFromFile',
-        file_client_args=file_client_args,
-        ignore_empty=True,
-        min_size=5),
-    dict(type='LoadOCRAnnotations', with_text=True),
-    dict(
-        type='RescaleToHeight',
-        height=32,
-        min_width=32,
-        max_width=160,
-        width_divisor=4),
-    dict(type='PadToWidth', width=160),
-    dict(
-        type='PackTextRecogInputs',
-        meta_keys=('img_path', 'ori_shape', 'img_shape', 'valid_ratio'))
-]
-
-test_pipeline = [
-    dict(type='LoadImageFromFile', file_client_args=file_client_args),
-    dict(
-        type='RescaleToHeight',
-        height=32,
-        min_width=32,
-        max_width=160,
-        width_divisor=16),
-    dict(type='PadToWidth', width=160),
-    # add loading annotation after ``Resize`` because ground truth
-    # does not need to do resize data transform
-    dict(type='LoadOCRAnnotations', with_text=True),
-    dict(
-        type='PackTextRecogInputs',
-        meta_keys=('img_path', 'ori_shape', 'img_shape', 'valid_ratio'))
-]
-
-train_dataloader = dict(
-    batch_size=384,
-    num_workers=32,
-    persistent_workers=True,
-    sampler=dict(type='DefaultSampler', shuffle=True),
-    dataset=dict(
-        type='ConcatDataset', datasets=train_list, pipeline=train_pipeline))
-
-test_dataloader = dict(
-    batch_size=1,
-    num_workers=4,
-    persistent_workers=True,
-    drop_last=False,
-    sampler=dict(type='DefaultSampler', shuffle=False),
-    dataset=dict(
-        type='ConcatDataset', datasets=test_list, pipeline=test_pipeline))
-val_dataloader = test_dataloader
-
-val_evaluator = dict(
-    type='MultiDatasetsEvaluator',
-    metrics=[
-        dict(
-            type='WordMetric',
-            mode=['exact', 'ignore_case', 'ignore_case_symbol']),
-        dict(type='CharMetric')
-    ],
-    dataset_prefixes=['CUTE80', 'IIIT5K', 'SVT', 'SVTP', 'IC13', 'IC15'])
-test_evaluator = val_evaluator
-
-visualizer = dict(type='TextRecogLocalVisualizer', name='visualizer')

configs/textrecog/nrtr/nrtr_modality_transform_toy_dataset.py

@@ -1,73 +0,0 @@
-_base_ = [
-    '../../_base_/recog_datasets/toy_data.py',
-    '../../_base_/default_runtime.py',
-    '../../_base_/schedules/schedule_adam_step_6e.py',
-    'nrtr_modality_transform.py',
-]
-
-# dataset settings
-train_list = {{_base_.train_list}}
-test_list = {{_base_.test_list}}
-file_client_args = dict(backend='disk')
-default_hooks = dict(logger=dict(type='LoggerHook', interval=50))
-
-train_pipeline = [
-    dict(type='LoadImageFromFile', file_client_args=file_client_args),
-    dict(type='LoadOCRAnnotations', with_text=True),
-    dict(
-        type='RescaleToHeight',
-        height=32,
-        min_width=32,
-        max_width=160,
-        width_divisor=4),
-    dict(type='PadToWidth', width=160),
-    dict(
-        type='PackTextRecogInputs',
-        meta_keys=('img_path', 'ori_shape', 'img_shape', 'valid_ratio'))
-]
-
-test_pipeline = [
-    dict(type='LoadImageFromFile', file_client_args=file_client_args),
-    dict(
-        type='RescaleToHeight',
-        height=32,
-        min_width=32,
-        max_width=160,
-        width_divisor=16),
-    dict(type='PadToWidth', width=160),
-    # add loading annotation after ``Resize`` because ground truth
-    # does not need to do resize data transform
-    dict(type='LoadOCRAnnotations', with_text=True),
-    dict(
-        type='PackTextRecogInputs',
-        meta_keys=('img_path', 'ori_shape', 'img_shape', 'valid_ratio'))
-]
-
-train_dataloader = dict(
-    batch_size=8,
-    num_workers=4,
-    persistent_workers=True,
-    sampler=dict(type='DefaultSampler', shuffle=True),
-    dataset=dict(
-        type='ConcatDataset', datasets=train_list, pipeline=train_pipeline))
-
-val_dataloader = dict(
-    batch_size=1,
-    num_workers=4,
-    persistent_workers=True,
-    drop_last=False,
-    sampler=dict(type='DefaultSampler', shuffle=False),
-    dataset=dict(
-        type='ConcatDataset', datasets=test_list, pipeline=test_pipeline))
-
-test_dataloader = val_dataloader
-
-val_evaluator = [
-    dict(
-        type='WordMetric', mode=['exact', 'ignore_case',
-                                 'ignore_case_symbol']),
-    dict(type='CharMetric')
-]
-
-test_evaluator = val_evaluator
-visualizer = dict(type='TextRecogLocalVisualizer', name='visualizer')

configs/textrecog/nrtr/nrtr_r31.py

@@ -1,30 +0,0 @@
-dictionary = dict(
-    type='Dictionary',
-    dict_file='dicts/english_digits_symbols.txt',
-    with_padding=True,
-    with_unknown=True,
-    same_start_end=True,
-    with_start=True,
-    with_end=True)
-
-model = dict(
-    type='NRTR',
-    backbone=dict(
-        type='ResNet31OCR',
-        layers=[1, 2, 5, 3],
-        channels=[32, 64, 128, 256, 512, 512],
-        stage4_pool_cfg=dict(kernel_size=(2, 1), stride=(2, 1)),
-        last_stage_pool=True),
-    encoder=dict(type='NRTREncoder'),
-    decoder=dict(
-        type='NRTRDecoder',
-        module_loss=dict(
-            type='CEModuleLoss', ignore_first_char=True, flatten=True),
-        postprocessor=dict(type='AttentionPostprocessor'),
-        dictionary=dictionary,
-        max_seq_len=30,
-    ),
-    data_preprocessor=dict(
-        type='TextRecogDataPreprocessor',
-        mean=[123.675, 116.28, 103.53],
-        std=[58.395, 57.12, 57.375]))

configs/textrecog/nrtr/nrtr_r31_1by16_1by8_academic.py

@@ -1,93 +0,0 @@
-_base_ = [
-    '../../_base_/recog_datasets/mjsynth.py',
-    '../../_base_/recog_datasets/synthtext.py',
-    '../../_base_/recog_datasets/cute80.py',
-    '../../_base_/recog_datasets/iiit5k.py',
-    '../../_base_/recog_datasets/svt.py',
-    '../../_base_/recog_datasets/svtp.py',
-    '../../_base_/recog_datasets/icdar2013.py',
-    '../../_base_/recog_datasets/icdar2015.py',
-    '../../_base_/default_runtime.py',
-    '../../_base_/schedules/schedule_adam_step_6e.py',
-    'nrtr_r31.py',
-]
-
-# optimizer settings
-optim_wrapper = dict(type='OptimWrapper', optimizer=dict(type='Adam', lr=3e-4))
-
-# dataset settings
-train_list = [_base_.mj_rec_train, _base_.st_rec_train]
-test_list = [
-    _base_.cute80_rec_test, _base_.iiit5k_rec_test, _base_.svt_rec_test,
-    _base_.svtp_rec_test, _base_.ic13_rec_test, _base_.ic15_rec_test
-]
-
-file_client_args = dict(backend='disk')
-default_hooks = dict(logger=dict(type='LoggerHook', interval=50))
-
-train_pipeline = [
-    dict(
-        type='LoadImageFromFile',
-        file_client_args=file_client_args,
-        ignore_empty=True,
-        min_size=5),
-    dict(type='LoadOCRAnnotations', with_text=True),
-    dict(
-        type='RescaleToHeight',
-        height=32,
-        min_width=32,
-        max_width=160,
-        width_divisor=4),
-    dict(type='PadToWidth', width=160),
-    dict(
-        type='PackTextRecogInputs',
-        meta_keys=('img_path', 'ori_shape', 'img_shape', 'valid_ratio'))
-]
-
-test_pipeline = [
-    dict(type='LoadImageFromFile', file_client_args=file_client_args),
-    dict(
-        type='RescaleToHeight',
-        height=32,
-        min_width=32,
-        max_width=160,
-        width_divisor=16),
-    dict(type='PadToWidth', width=160),
-    # add loading annotation after ``Resize`` because ground truth
-    # does not need to do resize data transform
-    dict(type='LoadOCRAnnotations', with_text=True),
-    dict(
-        type='PackTextRecogInputs',
-        meta_keys=('img_path', 'ori_shape', 'img_shape', 'valid_ratio'))
-]
-
-train_dataloader = dict(
-    batch_size=384,
-    num_workers=24,
-    persistent_workers=True,
-    sampler=dict(type='DefaultSampler', shuffle=True),
-    dataset=dict(
-        type='ConcatDataset', datasets=train_list, pipeline=train_pipeline))
-
-test_dataloader = dict(
-    batch_size=1,
-    num_workers=4,
-    persistent_workers=True,
-    drop_last=False,
-    sampler=dict(type='DefaultSampler', shuffle=False),
-    dataset=dict(
-        type='ConcatDataset', datasets=test_list, pipeline=test_pipeline))
-val_dataloader = test_dataloader
-
-val_evaluator = dict(
-    type='MultiDatasetsEvaluator',
-    metrics=[
-        dict(
-            type='WordMetric',
-            mode=['exact', 'ignore_case', 'ignore_case_symbol']),
-        dict(type='CharMetric')
-    ],
-    dataset_prefixes=['CUTE80', 'IIIT5K', 'SVT', 'SVTP', 'IC13', 'IC15'])
-test_evaluator = val_evaluator
-
-visualizer = dict(type='TextRecogLocalVisualizer', name='visualizer')

configs/textrecog/nrtr/nrtr_r31_1by8_1by4_academic.py

@@ -1,95 +0,0 @@
-_base_ = [
-    '../../_base_/recog_datasets/mjsynth.py',
-    '../../_base_/recog_datasets/synthtext.py',
-    '../../_base_/recog_datasets/cute80.py',
-    '../../_base_/recog_datasets/iiit5k.py',
-    '../../_base_/recog_datasets/svt.py',
-    '../../_base_/recog_datasets/svtp.py',
-    '../../_base_/recog_datasets/icdar2013.py',
-    '../../_base_/recog_datasets/icdar2015.py',
-    '../../_base_/default_runtime.py',
-    '../../_base_/schedules/schedule_adam_step_6e.py',
-    'nrtr_r31.py',
-]
-
-# optimizer settings
-optim_wrapper = dict(type='OptimWrapper', optimizer=dict(type='Adam', lr=3e-4))
-
-# dataset settings
-train_list = [_base_.mj_rec_train, _base_.st_rec_train]
-test_list = [
-    _base_.cute80_rec_test, _base_.iiit5k_rec_test, _base_.svt_rec_test,
-    _base_.svtp_rec_test, _base_.ic13_rec_test, _base_.ic15_rec_test
-]
-
-file_client_args = dict(backend='disk')
-default_hooks = dict(logger=dict(type='LoggerHook', interval=50), )
-
-model = dict(backbone=dict(last_stage_pool=False))
-
-train_pipeline = [
-    dict(
-        type='LoadImageFromFile',
-        file_client_args=file_client_args,
-        ignore_empty=True,
-        min_size=5),
-    dict(type='LoadOCRAnnotations', with_text=True),
-    dict(
-        type='RescaleToHeight',
-        height=32,
-        min_width=32,
-        max_width=160,
-        width_divisor=4),
-    dict(type='PadToWidth', width=160),
-    dict(
-        type='PackTextRecogInputs',
-        meta_keys=('img_path', 'ori_shape', 'img_shape', 'valid_ratio'))
-]
-
-test_pipeline = [
-    dict(type='LoadImageFromFile', file_client_args=file_client_args),
-    dict(
-        type='RescaleToHeight',
-        height=32,
-        min_width=32,
-        max_width=160,
-        width_divisor=16),
-    dict(type='PadToWidth', width=160),
-    # add loading annotation after ``Resize`` because ground truth
-    # does not need to do resize data transform
-    dict(type='LoadOCRAnnotations', with_text=True),
-    dict(
-        type='PackTextRecogInputs',
-        meta_keys=('img_path', 'ori_shape', 'img_shape', 'valid_ratio'))
-]
-
-train_dataloader = dict(
-    batch_size=384,
-    num_workers=32,
-    persistent_workers=True,
-    sampler=dict(type='DefaultSampler', shuffle=True),
-    dataset=dict(
-        type='ConcatDataset', datasets=train_list, pipeline=train_pipeline))
-
-test_dataloader = dict(
-    batch_size=1,
-    num_workers=4,
-    persistent_workers=True,
-    drop_last=False,
-    sampler=dict(type='DefaultSampler', shuffle=False),
-    dataset=dict(
-        type='ConcatDataset', datasets=test_list, pipeline=test_pipeline))
-val_dataloader = test_dataloader
-
-val_evaluator = dict(
-    type='MultiDatasetsEvaluator',
-    metrics=[
-        dict(
-            type='WordMetric',
-            mode=['exact', 'ignore_case', 'ignore_case_symbol']),
-        dict(type='CharMetric')
-    ],
-    dataset_prefixes=['CUTE80', 'IIIT5K', 'SVT', 'SVTP', 'IC13', 'IC15'])
-test_evaluator = val_evaluator
-
-visualizer = dict(type='TextRecogLocalVisualizer', name='visualizer')

configs/textrecog/nrtr/nrtr_resnet31-1by16-1by8_6e_st_mj.py

@@ -0,0 +1,45 @@
+_base_ = [
+    '../../_base_/recog_datasets/mjsynth.py',
+    '../../_base_/recog_datasets/synthtext.py',
+    '../../_base_/recog_datasets/cute80.py',
+    '../../_base_/recog_datasets/iiit5k.py',
+    '../../_base_/recog_datasets/svt.py',
+    '../../_base_/recog_datasets/svtp.py',
+    '../../_base_/recog_datasets/icdar2013.py',
+    '../../_base_/recog_datasets/icdar2015.py',
+    '../../_base_/textrec_default_runtime.py',
+    '../../_base_/schedules/schedule_adam_step_6e.py',
+    '_base_nrtr_resnet31.py',
+]
+
+# optimizer settings
+optim_wrapper = dict(type='OptimWrapper', optimizer=dict(type='Adam', lr=3e-4))
+
+# dataset settings
+train_list = [_base_.mj_rec_train, _base_.st_rec_train]
+test_list = [
+    _base_.cute80_rec_test, _base_.iiit5k_rec_test, _base_.svt_rec_test,
+    _base_.svtp_rec_test, _base_.ic13_rec_test, _base_.ic15_rec_test
+]
+
+train_dataset = dict(
+    type='ConcatDataset', datasets=train_list, pipeline=_base_.train_pipeline)
+test_dataset = dict(
+    type='ConcatDataset', datasets=test_list, pipeline=_base_.test_pipeline)
+
+train_dataloader = dict(
+    batch_size=384,
+    num_workers=24,
+    persistent_workers=True,
+    sampler=dict(type='DefaultSampler', shuffle=True),
+    dataset=train_dataset)
+
+test_dataloader = dict(
+    batch_size=1,
+    num_workers=4,
+    persistent_workers=True,
+    drop_last=False,
+    sampler=dict(type='DefaultSampler', shuffle=False),
+    dataset=test_dataset)
+
+val_dataloader = test_dataloader

configs/textrecog/nrtr/nrtr_resnet31-1by8-1by4_6e_st_mj.py

@@ -0,0 +1,5 @@
+_base_ = [
+    'nrtr_resnet31-1by16-1by8_6e_st_mj.py',
+]
+
+model = dict(backbone=dict(last_stage_pool=False))