Update app.py
Browse files
app.py
CHANGED
@@ -83,8 +83,8 @@ class TrainingConfig:
|
|
83 |
@dataclass(frozen=True)
|
84 |
class DatasetConfig:
|
85 |
DATA_ROOT: str = 'DATASET-TRAIN-TEST-1'
|
86 |
-
TRAIN_DATA_ROOT: str = 'filetxt/train/
|
87 |
-
|
88 |
@dataclass(frozen=True)
|
89 |
class ModelConfig:
|
90 |
MODEL_NAME: str = 'microsoft/trocr-small-printed'
|
@@ -104,11 +104,11 @@ def visualize(dataset_path):
|
|
104 |
visualize(DatasetConfig.DATA_ROOT)
|
105 |
|
106 |
train_df = pd.read_fwf(
|
107 |
-
os.path.join(DatasetConfig.TRAIN_DATA_ROOT, '
|
108 |
)
|
109 |
train_df.rename(columns={0: 'file_name', 1: 'text'}, inplace=True)
|
110 |
test_df = pd.read_fwf(
|
111 |
-
os.path.join(DatasetConfig.
|
112 |
)
|
113 |
test_df.rename(columns={0: 'file_name', 1: 'text'}, inplace=True)
|
114 |
|
|
|
83 |
@dataclass(frozen=True)
|
84 |
class DatasetConfig:
|
85 |
DATA_ROOT: str = 'DATASET-TRAIN-TEST-1'
|
86 |
+
TRAIN_DATA_ROOT: str = 'filetxt/DATASET TXT/train/'
|
87 |
+
TEST_DATA_ROOT: str = 'filetxt/DATASET TXT/test/'
|
88 |
@dataclass(frozen=True)
|
89 |
class ModelConfig:
|
90 |
MODEL_NAME: str = 'microsoft/trocr-small-printed'
|
|
|
104 |
visualize(DatasetConfig.DATA_ROOT)
|
105 |
|
106 |
train_df = pd.read_fwf(
|
107 |
+
os.path.join(DatasetConfig.TRAIN_DATA_ROOT, 'train.txt'), header=None
|
108 |
)
|
109 |
train_df.rename(columns={0: 'file_name', 1: 'text'}, inplace=True)
|
110 |
test_df = pd.read_fwf(
|
111 |
+
os.path.join(DatasetConfig.TEST_DATA_ROOT, 'test.txt'), header=None
|
112 |
)
|
113 |
test_df.rename(columns={0: 'file_name', 1: 'text'}, inplace=True)
|
114 |
|