How to capture data from tables

Who can use this feature:

  • Pro, Enterprise : Custom models as well as Pre-built models (eg: Invoices, Receipts, etc) will have access to this feature on the paid plans only. Anyone on your Team with can edit access can capture and label data from Tables.

Capture data from tables (Video) 

Please follow the steps in the video or check out the steps listed below for written instructions.

Capture data from tables (Instructions) 

  1. Draw a regular box over the table on your image.
  2. From the label popup > Switch to Tables using the toggle button on the top left.
  3. You will see a selector for rows and columns. Select the number of rows and columns you need.
  4. The box you drew will now be replaced by a grid.
  5. To adjust the gridlines to match the rows and columns on your image > click and drag on a gridline.
  6. To adjust the overall size > click and drag on the corners of the gird.
  7. Click on Extract data as table.
  8. You will see the data captured in tabular format under the Tables section on the panel to the right.
  9. You will see label bounding boxes added for each cell on your image.
  10. Follow the steps below to edit table data.

Edit table data

Once you've captured the data from the grid, you can edit the boxes such that you extract only the data you need. For example: Your product column contains Product_code and Product_description but you only need to extract Product_code.

Edit data using the boxes on the image:

  1. On the image > click on the box for the cell data you want to edit.
  2. Drag to Resize the box so that it contains only the data you need.
  3. You will see the data reflect on the panel to the right.


Edit data from the cells on the panel to the right:

  1. On the panel to the right > Click on the cell you want to edit.
  2. Type to edit the text or select the text and hit delete to clear.
    Please note: The bounding box on the image will not auto-resize to match the data you manually edit.

Assign Column name labels to table data


You do not ned to assign the same label name as the text for that column on your image. For example: If your image has a column "Product code" but your database uses "Item ID", you can label the extracted column Item ID. Nanonets export will show the column name you add.

To assign column names or headers to the extracted data, follow the steps to assign a label from the panel to the right.

  1. Find the column you need to label.
  2. On the top of each column you will see a dropdown and input field (Header)
  3. Type the column name you want to create.
  4. Click on Add <your column name> 
  5. All the cells below that header will be assigned to that column label.

Once you have added the column names you need, you can simply select from the name dropdown on the other images.


When training a custom model with Tables, you need to add at least 10 examples for each column name you add. For example: If you have added the column label "Quantity" on one image, you will need to mark 10 other columns on any other images with the name "Quantity".

How do I hide the gridlines on the image?

On the bottom bar under the image > click on Hide grids.

Or, to hide a grid for a particular table only:

  1. Click on the dropdown icon next to a Table name on the panel to the right
  2. Select Hide grid. (This will hide the grid only for the Table you selected)

How do I delete extra/junk boxes captured?

Though we currently don't support deleting of boxes directly from the image, you can select the data on the cell and clear it.

Was this article helpful?