BritishRainfall

Table Cells and Words identified by Amazon Textract within images of Rainfall Sheets

Using Textract in ‘Analyze Document’ mode to detect Table features as well as text.

• DR_Anglesey_1871-1880C_p002
• DR_Anglesey_1871-1880C_p005
• DR_Anglesey_1871-1880C_p008
• DR_Anglesey_1871-1880C_p017
• DR_Anglesey_1871-1880C_p020

Blue and black rectangles on the images show the bounding boxes for each ‘Word’ identified by Amazon Textract. Click within a rectangle to see a pop-up showing the text which Amazon Textract detected.

Yellow and red rectangles show the bounding boxes for table cells. Click within a rectangle to see a pop-up identifying the row/column of the cell, and the text of the ‘Word’ associated with the cell. Cells with a red border have more than one ‘Word’ associated with them.

(You can close the pop-ups with the ‘Esc’ key as well as by pressing the OK button.)

An example where Textract table detection hasn’t worked properly for determining the rows in the table:

• DR_Anglesey_1871-1880C_p004 - the rows determined by Textract gradually get out of alignment with the actual rows of figures in the image