-
Notifications
You must be signed in to change notification settings - Fork 44
/
rare_resnet34_ch.yaml
156 lines (144 loc) · 3.6 KB
/
rare_resnet34_ch.yaml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
system:
mode: 0 # 0 for graph mode, 1 for pynative mode in MindSpore
distribute: True
amp_level: O2
amp_level_infer: O2 # running inference in O2 mode
seed: 42
log_interval: 100
val_while_train: True
drop_overflow_update: False
ckpt_max_keep: 5
common:
character_dict_path: &character_dict_path mindocr/utils/dict/ch_dict.txt
num_classes: &num_classes 6625 # num_chars_in_dict + 2
max_text_len: &max_text_len 40
infer_mode: &infer_mode False
use_space_char: &use_space_char False
batch_size: &batch_size 256
model:
type: rec
transform: null
backbone:
name: rec_resnet34
pretrained: False
neck:
name: RNNEncoder
hidden_size: 64
head:
name: AttentionHead
hidden_size: 256
out_channels: *num_classes
batch_max_length: *max_text_len
postprocess:
name: RecAttnLabelDecode
character_dict_path: *character_dict_path
use_space_char: *use_space_char
metric:
name: RecMetric
main_indicator: acc
lower: False
character_dict_path: *character_dict_path
ignore_space: True
print_flag: False
loss:
name: AttentionLoss
scheduler:
scheduler: warmup_cosine_decay
min_lr: 0.00001
lr: 0.001
num_epochs: 60
warmup_epochs: 6
decay_epochs: 54
optimizer:
opt: adamw
filter_bias_and_bn: True
weight_decay: 0.05
loss_scaler:
type: dynamic
loss_scale: 512
scale_factor: 2.0
scale_window: 1000
train:
ckpt_save_dir: ./tmp_rec
dataset_sink_mode: False
clip_grad: True
clip_norm: 0.1
dataset:
type: LMDBDataset
dataset_root: path/to/chinese_text_recognition/
data_dir: training/
label_file: null
sample_ratio: 1.0
shuffle: True
filter_max_len: True
max_text_len: *max_text_len
transform_pipeline:
- DecodeImage:
img_mode: BGR
to_float32: False
- RecAttnLabelEncode:
max_text_len: *max_text_len
character_dict_path: *character_dict_path
use_space_char: *use_space_char
lower: False
- Rotate90IfVertical:
threshold: 2.0
direction: counterclockwise
- RecResizeImg:
image_shape: [32, 320]
infer_mode: *infer_mode
character_dict_path: *character_dict_path
padding: True
- NormalizeImage:
bgr_to_rgb: True
is_hwc: True
mean: [127.0, 127.0, 127.0]
std: [127.0, 127.0, 127.0]
- ToCHWImage:
output_columns: ["image", "text_seq"]
net_input_column_index: [0, 1]
label_column_index: [1]
loader:
shuffle: True
batch_size: *batch_size
drop_remainder: True
max_rowsize: 12
num_workers: 1
eval:
ckpt_load_path: ./tmp_rec/best.ckpt
dataset_sink_mode: False
dataset:
type: LMDBDataset
dataset_root: path/to/chinese_text_recognition/
data_dir: validation/
label_file: null
sample_ratio: 1.0
shuffle: False
transform_pipeline:
- DecodeImage:
img_mode: RGB
to_float32: False
- RecAttnLabelEncode:
max_text_len: *max_text_len
character_dict_path: *character_dict_path
use_space_char: *use_space_char
lower: False
- Rotate90IfVertical:
threshold: 2.0
direction: counterclockwise
- RecResizeNormForInfer:
target_height: 32
target_width: 320
keep_ratio: True
padding: True
norm_before_pad: False
- ToCHWImage:
output_columns: ["image", "text_padded", "text_length"]
net_input_column_index: [0]
label_column_index: [1, 2]
loader:
shuffle: False
batch_size: 64
drop_remainder: False
max_rowsize: 12
num_workers: 1