Spaces:
Runtime error
Runtime error
Upload 2 files
Browse files- datatypes/config.py +13 -0
- datatypes/datatypes.py +53 -0
datatypes/config.py
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
|
| 3 |
+
# config values
|
| 4 |
+
|
| 5 |
+
Config = {"table_detection_padding_pixel":10, 'table_recognition_padding_pixel':5,
|
| 6 |
+
'table_detection_threshold':0.7, 'table_recognition_threshold':0.8,
|
| 7 |
+
'table_padd': 20, 'row_padd':6, 'cell_padd':3,
|
| 8 |
+
}
|
| 9 |
+
|
| 10 |
+
tesseract_config = {'tesseractpath':'C://Program Files//Tesseract-OCR//tesseract.exe'}
|
| 11 |
+
|
| 12 |
+
model_config = {'detection_model_path':'D:/Table-detection/models/detection-model',
|
| 13 |
+
'recognition_model_path':'D:/Table-detection/models/recognition-model'}
|
datatypes/datatypes.py
ADDED
|
@@ -0,0 +1,53 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
from dataclasses import dataclass, field
|
| 3 |
+
from enum import Enum
|
| 4 |
+
from typing import List, Dict
|
| 5 |
+
|
| 6 |
+
class DetectionLabels(Enum):
|
| 7 |
+
table = 0
|
| 8 |
+
table_column = 1
|
| 9 |
+
table_row = 2
|
| 10 |
+
table_column_header = 3
|
| 11 |
+
table_projected_row_header = 4
|
| 12 |
+
table_spanning_cell = 5
|
| 13 |
+
|
| 14 |
+
class ExtractionContext(Enum):
|
| 15 |
+
document = 1
|
| 16 |
+
table = 2
|
| 17 |
+
row = 3
|
| 18 |
+
|
| 19 |
+
@dataclass
|
| 20 |
+
class Cell:
|
| 21 |
+
cellindex : int = 0
|
| 22 |
+
value : str = ''
|
| 23 |
+
prob : float = 0.5
|
| 24 |
+
|
| 25 |
+
@dataclass
|
| 26 |
+
class Row:
|
| 27 |
+
rowindex : int = 0
|
| 28 |
+
extracted_cells : List[Cell]= field(default_factory=lambda: [])
|
| 29 |
+
|
| 30 |
+
@dataclass
|
| 31 |
+
class TableRecognitionData:
|
| 32 |
+
scores : List = field(default_factory=lambda: [])
|
| 33 |
+
labels : List = field(default_factory=lambda: [])
|
| 34 |
+
boxes : List = field(default_factory=lambda: [])
|
| 35 |
+
|
| 36 |
+
@dataclass
|
| 37 |
+
class TableRecognitionOrdered:
|
| 38 |
+
recognized_row : List = field(default_factory=lambda: [])
|
| 39 |
+
recognized_column : List = field(default_factory=lambda: [])
|
| 40 |
+
|
| 41 |
+
@dataclass
|
| 42 |
+
class TableDetectionData:
|
| 43 |
+
detection_score : float = 0.0
|
| 44 |
+
detection_label : int = 0
|
| 45 |
+
detection_box : List = field(default_factory=lambda: [])
|
| 46 |
+
recognitiondata : TableRecognitionData = field(default_factory=lambda: [])
|
| 47 |
+
ordered_recognitiondata : List[TableRecognitionOrdered] = field(default_factory=lambda: [])
|
| 48 |
+
extracted_rows : List[Row] = field(default_factory=lambda: [])
|
| 49 |
+
|
| 50 |
+
|
| 51 |
+
@dataclass
|
| 52 |
+
class ImageData:
|
| 53 |
+
tables: List[TableDetectionData]
|