TableFormer

TableFormer with json and accurate coordinates.¶

refer: docs/getting_started/table_test.ipynb

In [1]:

from omnidocs.tasks.table_extraction.extractors import TableFormerExtractor

from omnidocs.tasks.table_extraction.extractors import TableFormerExtractor

c:\Users\laxma\OneDrive\Desktop\CogLab\11-07-2025\Omnidocs\new\Lib\site-packages\transformers\utils\hub.py:111: FutureWarning: Using `TRANSFORMERS_CACHE` is deprecated and will be removed in v5 of Transformers. Use `HF_HOME` instead.
  warnings.warn(

In [2]:

image_path = "../../../../tests/table_extraction/assets/table_image.png"
result = TableFormerExtractor().extract(image_path)

image_path = "../../../../tests/table_extraction/assets/table_image.png" result = TableFormerExtractor().extract(image_path)

INFO     [timestamp]2025-07-31 13:29:40[/] | [logger.name]omnidocs.tasks.table_extraction.extractors.tableformer[/]
         | [function]logging.py:150[/] | [info]extract completed in 1.50s[/]                                       

INFO     [timestamp]2025-07-31 13:29:40[/] | [logger.name]omnidocs.tasks.table_extraction.extractors.tableformer[/]
         | [function]logging.py:150[/] | [info]extract completed in 1.50s[/]                                       

In [3]:

for i, table in enumerate(result.tables):
                print(f"\nTable {i+1}: {table.num_rows} rows x {table.num_cols} columns")
                print(f"Total cells: {len(table.cells)}")

                if table.cells:
                    non_empty_cells = [cell for cell in table.cells if cell.text.strip()]
                    print(f"Non-empty cells: {len(non_empty_cells)}")

                    # Show first few cells
                    for cell in table.cells[:10]:
                        if cell.text.strip():
                            text = cell.text.strip()[:30]
                            print(f"  [{cell.row},{cell.col}]: '{text}'")

for i, table in enumerate(result.tables): print(f"\nTable {i+1}: {table.num_rows} rows x {table.num_cols} columns") print(f"Total cells: {len(table.cells)}") if table.cells: non_empty_cells = [cell for cell in table.cells if cell.text.strip()] print(f"Non-empty cells: {len(non_empty_cells)}") # Show first few cells for cell in table.cells[:10]: if cell.text.strip(): text = cell.text.strip()[:30] print(f" [{cell.row},{cell.col}]: '{text}'")

Table 1: 6 rows x 7 columns
Total cells: 39
Non-empty cells: 39
  [0,0]: 'Cell_0_0'
  [0,1]: 'Cell_0_1'
  [0,2]: 'Cell_0_2'
  [0,3]: 'Cell_0_3'
  [0,4]: 'Cell_0_4'
  [0,5]: 'Cell_0_5'
  [0,6]: 'Cell_0_6'
  [1,0]: 'Cell_1_0'
  [1,1]: 'Cell_1_1'
  [1,2]: 'Cell_1_2'

In [ ]: