4x-tech/easyocr
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

testing dataset ok; trainer.py take effects

Browse Source
This commit is contained in:
HT
2025-07-11 11:08:51 +08:00
parent 185959cf2a
commit cfa52e5f2e
11 changed files with 167 additions and 97 deletions
Download Patch File Download Diff File Expand all files Collapse all files

46
trainer/config_files/4digit_config.yaml Normal file
View File

@@ -0,0 +1,46 @@
number: '0123456789'
experiment_name: '4digit'
symbol: ""
lang_char: ''
train_data: 'all_data'
valid_data: 'all_data/4digit_valid'
manualSeed: 1111
workers: 6
batch_size: 32 #32
num_iter: 3000
valInterval: 5
# saved_model: '' #'saved_models/en_filtered/iter_300000.pth'
svaed_model: 'saved_models/4digit/iter_3000.pth'
FT: False
optim: False # default is Adadelta
lr: 1.
beta1: 0.9
rho: 0.95
eps: 0.00000001
grad_clip: 5
#Data processing
select_data: '4digit_train' # this is dataset folder in train_data
batch_ratio: '1'
total_data_usage_ratio: 1.0
batch_max_length: 34
imgH: 32
imgW: 128
rgb: True
contrast_adjust: False
sensitive: True
PAD: True
contrast_adjust: 0.0
data_filtering_off: False
# Model Architecture
Transformation: 'TPS'
FeatureExtraction: 'ResNet'
SequenceModeling: 'BiLSTM'
Prediction: 'CTC'
num_fiducial: 20
input_channel: 1
output_channel: 256
hidden_size: 256
decode: 'greedy'
new_prediction: False
freeze_FeatureFxtraction: False
freeze_SequenceModeling: False

8
trainer/config_files/en_filtered_config.yaml
View File

@@ -7,8 +7,8 @@ valid_data: 'all_data/valid'
manualSeed: 1111
workers: 6
batch_size: 32 #32
num_iter: 300000
valInterval: 20000
num_iter: 300
valInterval: 5
saved_model: '' #'saved_models/en_filtered/iter_300000.pth'
FT: False
optim: False # default is Adadelta
@@ -31,8 +31,8 @@ PAD: True
contrast_adjust: 0.0
data_filtering_off: False
# Model Architecture
Transformation: 'ResNet'
FeatureExtraction: 'VGG'
Transformation: 'None'
FeatureExtraction: 'ResNet'
SequenceModeling: 'BiLSTM'
Prediction: 'CTC'
num_fiducial: 20

14
trainer/dataset.py
View File

@@ -42,7 +42,6 @@ class Batch_Balanced_Dataset(object):
log.write(dashed_line + '\n')
print(f'dataset_root: {opt.train_data}\nopt.select_data: {opt.select_data}\nopt.batch_ratio: {opt.batch_ratio}')
log.write(f'dataset_root: {opt.train_data}\nopt.select_data: {opt.select_data}\nopt.batch_ratio: {opt.batch_ratio}\n')
print(f"len(opt.select_data): {len(opt.select_data)}, len(opt.batch_ratio): {len(opt.batch_ratio)}")
assert len(opt.select_data) == len(opt.batch_ratio)
_AlignCollate = AlignCollate(imgH=opt.imgH, imgW=opt.imgW, keep_ratio_with_pad=opt.PAD, contrast_adjust = opt.contrast_adjust)
@@ -54,7 +53,6 @@ class Batch_Balanced_Dataset(object):
_batch_size = max(round(opt.batch_size * float(batch_ratio_d)), 1)
print(dashed_line)
log.write(dashed_line + '\n')
print(f"selected_d: {selected_d}, batch_ratio: {batch_ratio_d}, batch_size: {_batch_size}")
_dataset, _dataset_log = hierarchical_dataset(root=opt.train_data, opt=opt, select_data=[selected_d])
total_number_dataset = len(_dataset)
log.write(_dataset_log)
@@ -100,12 +98,12 @@ class Batch_Balanced_Dataset(object):
for i, data_loader_iter in enumerate(self.dataloader_iter_list):
try:
image, text = data_loader_iter.next()
image, text = next(data_loader_iter)
balanced_batch_images.append(image)
balanced_batch_texts += text
except StopIteration:
self.dataloader_iter_list[i] = iter(self.data_loader_list[i])
image, text = self.dataloader_iter_list[i].next()
image, text = next(self.dataloader_iter_list[i])
balanced_batch_images.append(image)
balanced_batch_texts += text
except ValueError:
@@ -119,7 +117,7 @@ class Batch_Balanced_Dataset(object):
def hierarchical_dataset(root, opt, select_data='/'):
""" select_data='/' contains all sub-directory of root directory """
dataset_list = []
dataset_log = f'dataset_root: {root}\t dataset[0]: {select_data[0]}'
dataset_log = f'dataset_root: {root}\t dataset: {select_data[0]}'
print(dataset_log)
dataset_log += '\n'
for dirpath, dirnames, filenames in os.walk(root+'/'):
@@ -148,12 +146,8 @@ class OCRDataset(Dataset):
self.root = root
self.opt = opt
print(root)
print(f"Loading dataset from {root}...")
print(opt)
self.df = pd.read_csv(os.path.join(root,'labels.csv'), sep='^([^,]+),', engine='python', usecols=['filename', 'words'], keep_default_na=False)
self.nSamples = len(self.df)
print(f"Number of samples: {self.nSamples}")
if self.opt.data_filtering_off:
self.filtered_index_list = [index + 1 for index in range(self.nSamples)]
@@ -286,4 +280,4 @@ def tensor2im(image_tensor, imtype=np.uint8):
def save_image(image_numpy, image_path):
image_pil = Image.fromarray(image_numpy)
image_pil.save(image_path)
image_pil.save(image_path)

BIN
trainer/saved_models/en_filtered/best_accuracy.pth Normal file
View File

Binary file not shown.

BIN
trainer/saved_models/en_filtered/best_norm_ED.pth Normal file
View File

Binary file not shown.

88
trainer/saved_models/en_filtered/log_dataset.txt
View File

@@ -1,102 +1,30 @@
--------------------------------------------------------------------------------
dataset_root: all_data/en_sample
opt.select_data: ['all_data/en_sample/train']
opt.batch_ratio: ['1']
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
dataset_root: all_data/en_sample
opt.select_data: ['all_data/en_sample/train']
opt.batch_ratio: ['1']
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
dataset_root: all_data/en_sample
opt.select_data: ['all_data/en_sample/train']
opt.batch_ratio: ['1']
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
dataset_root: all_data/en_sample
opt.select_data: ['all_data/en_sample/train']
opt.batch_ratio: ['1']
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
dataset_root: all_data/en_sample
opt.select_data: ['all_data/en_sample/train']
opt.batch_ratio: ['1']
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
dataset_root: all_data/en_sample
opt.select_data: ['all_data/en_sample/train']
opt.batch_ratio: ['1']
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
dataset_root: all_data/
dataset_root: all_data
opt.select_data: ['train']
opt.batch_ratio: ['1']
--------------------------------------------------------------------------------
dataset_root: all_data/ dataset[0]: train
dataset_root: all_data dataset: train
sub-directory: /train num samples: 688
num total samples of train: 688 x 1.0 (total_data_usage_ratio) = 688
num samples of train per batch: 32 x 1.0 (batch_ratio) = 32
--------------------------------------------------------------------------------
Total_batch_size: 32 = 32
--------------------------------------------------------------------------------
dataset_root: all_data/valid dataset: /
sub-directory: /. num samples: 194
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
dataset_root: all_data
opt.select_data: ['train']
opt.batch_ratio: ['1']
--------------------------------------------------------------------------------
dataset_root: all_data dataset[0]: train
dataset_root: all_data dataset: train
sub-directory: /train num samples: 688
num total samples of train: 688 x 1.0 (total_data_usage_ratio) = 688
num samples of train per batch: 32 x 1.0 (batch_ratio) = 32
--------------------------------------------------------------------------------
Total_batch_size: 32 = 32
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
dataset_root: all_data
opt.select_data: ['train']
opt.batch_ratio: ['1']
--------------------------------------------------------------------------------
dataset_root: all_data dataset[0]: train
sub-directory: /train num samples: 688
num total samples of train: 688 x 1.0 (total_data_usage_ratio) = 688
num samples of train per batch: 32 x 1.0 (batch_ratio) = 32
--------------------------------------------------------------------------------
Total_batch_size: 32 = 32
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
dataset_root: all_data
opt.select_data: ['train']
opt.batch_ratio: ['1']
--------------------------------------------------------------------------------
dataset_root: all_data dataset[0]: train
sub-directory: /train num samples: 688
num total samples of train: 688 x 1.0 (total_data_usage_ratio) = 688
num samples of train per batch: 32 x 1.0 (batch_ratio) = 32
--------------------------------------------------------------------------------
Total_batch_size: 32 = 32
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
dataset_root: all_data
opt.select_data: ['train']
opt.batch_ratio: ['1']
--------------------------------------------------------------------------------
dataset_root: all_data dataset[0]: train
sub-directory: /train num samples: 688
num total samples of train: 688 x 1.0 (total_data_usage_ratio) = 688
num samples of train per batch: 32 x 1.0 (batch_ratio) = 32
--------------------------------------------------------------------------------
Total_batch_size: 32 = 32
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
dataset_root: all_data
opt.select_data: ['train']
opt.batch_ratio: ['1']
--------------------------------------------------------------------------------
dataset_root: all_data dataset[0]: train
sub-directory: /train num samples: 688
num total samples of train: 688 x 1.0 (total_data_usage_ratio) = 688
num samples of train per batch: 32 x 1.0 (batch_ratio) = 32
--------------------------------------------------------------------------------
Total_batch_size: 32 = 32
dataset_root: all_data/valid dataset: /
sub-directory: /. num samples: 194
--------------------------------------------------------------------------------

9
trainer/saved_models/en_filtered/log_train.txt Normal file
View File

@@ -0,0 +1,9 @@
[5/300] Train loss: 28.86501, Valid loss: 21.44183, Elapsed_time: 3.73978
Current_accuracy : 0.000, Current_norm_ED : 0.0445
Best_accuracy : 0.000, Best_norm_ED : 0.0445
--------------------------------------------------------------------------------
Ground Truth | Prediction | Confidence Score & T/F
--------------------------------------------------------------------------------
"Karjalan outcrop ""Sven""" | eee | 0.0000 False
ESPN Modernize | ee | 0.0000 False
--------------------------------------------------------------------------------

92
trainer/saved_models/en_filtered/opt.txt Normal file
View File

@@ -0,0 +1,92 @@
------------ Options -------------
number: 0123456789
symbol: !"#$%&'()*+,-./:;<=>?@[\]^_`{|}~ €
lang_char: ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz
experiment_name: en_filtered
train_data: all_data
valid_data: all_data/valid
manualSeed: 1111
workers: 6
batch_size: 32
num_iter: 300
valInterval: 5
saved_model:
FT: False
optim: False
lr: 1.0
beta1: 0.9
rho: 0.95
eps: 1e-08
grad_clip: 5
select_data: ['train']
batch_ratio: ['1']
total_data_usage_ratio: 1.0
batch_max_length: 34
imgH: 64
imgW: 600
rgb: False
contrast_adjust: 0.0
sensitive: True
PAD: True
data_filtering_off: False
Transformation: None
FeatureExtraction: ResNet
SequenceModeling: BiLSTM
Prediction: CTC
num_fiducial: 20
input_channel: 1
output_channel: 256
hidden_size: 256
decode: greedy
new_prediction: False
freeze_FeatureFxtraction: False
freeze_SequenceModeling: False
character: 0123456789!"#$%&'()*+,-./:;<=>?@[\]^_`{|}~ €ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz
num_class: 97
---------------------------------------
------------ Options -------------
number: 0123456789
symbol: !"#$%&'()*+,-./:;<=>?@[\]^_`{|}~ €
lang_char: ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz
experiment_name: en_filtered
train_data: all_data
valid_data: all_data/valid
manualSeed: 1111
workers: 6
batch_size: 32
num_iter: 300
valInterval: 5
saved_model:
FT: False
optim: False
lr: 1.0
beta1: 0.9
rho: 0.95
eps: 1e-08
grad_clip: 5
select_data: ['train']
batch_ratio: ['1']
total_data_usage_ratio: 1.0
batch_max_length: 34
imgH: 64
imgW: 600
rgb: False
contrast_adjust: 0.0
sensitive: True
PAD: True
data_filtering_off: False
Transformation: None
FeatureExtraction: ResNet
SequenceModeling: BiLSTM
Prediction: CTC
num_fiducial: 20
input_channel: 1
output_channel: 256
hidden_size: 256
decode: greedy
new_prediction: False
freeze_FeatureFxtraction: False
freeze_SequenceModeling: False
character: 0123456789!"#$%&'()*+,-./:;<=>?@[\]^_`{|}~ €ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz
num_class: 97
---------------------------------------

2
trainer/train.py
View File

@@ -68,7 +68,7 @@ def train(opt, show_number = 2, amp=False):
model = Model(opt)
print('model input parameters', opt.imgH, opt.imgW, opt.num_fiducial, opt.input_channel, opt.output_channel,
opt.hidden_size, opt.num_class, opt.batch_max_length, opt.Transformation, opt.FeatureExtraction,
opt.SequenceModeling, opt.Prediction)
opt.SequenceModeling, opt.Prediction, opt.saved_model)
if opt.saved_model != '':
pretrained_dict = torch.load(opt.saved_model)

4
trainer/trainer.ipynb
View File

@@ -93,7 +93,7 @@
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"display_name": "easyocr",
"language": "python",
"name": "python3"
},
@@ -107,7 +107,7 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.8.11"
"version": "3.8.20"
}
},
"nbformat": 4,

1
trainer/trainer.py
View File

@@ -26,5 +26,6 @@ def get_config(file_path):
if __name__ == "__main__":
# Load configuration
# opt = get_config("config_files/4digit_config.yaml")
opt = get_config("config_files/en_filtered_config.yaml")
train(opt, amp=False)