Recognize the table and include the title in the table

[Lihwnlp]

Very helpful project！！
I have a question: when I use .extract_tables to extract tables, the table's title and the text below the table are also being recognized. How should I configure it?
My form looks like this:

The extracted table looks like this；

[[[‘Table 2: abcdefg, None,None,None],
[‘ss', ‘ss’, ‘ss', ‘ss’],
[‘ss', ‘ss’, ‘ss', ‘ss’],
[‘ss', ‘ss’, ‘ss', ‘ss’],
[‘ss', ‘ss’, ‘ss', ‘ss’],
['4.9. cwniocnoiwenciewnciowenciewnc,None,None,None,None]]]

[jsvine]

Hi @Lihwnlp, can you attach the PDF in question? Without it, it will be difficult to propose a specific solution for your specific situation. Based on the screenshots, however, it seems that there are likely some non-visible graphical elements being picked up (perhaps because they are white-on-white). You can check whether this is the case by running page.to_image().debug_tablefinder().

You can also examine the .rects, .lines, and .curves object lists to try to figure out which specific objects are causing the issue, and then use page.filter(...) to exclude them. (Searching for filter in the discussions should point you to solutions for similar situations.)