Forum Discussion
Extract table from image using copilot
Hi everyone,
I could need some helo 🙂
How do I use Copilot to copy a table from an image? The table is in a larger PDF and I want to copy it into Copilot using the screenshot function.
The table can look like this. For each row in column A, there can be several columns in B.
I can only do it if I give the AI so much information, e.g. what the rows in column A are called. I use Copilot which is for business (I think only the data protection is different here).
This is a part of a bigger task. But it already fails to read in the table. Even when I tell it to use OCR.
The table can be like this example. A is a merged cell. Its not two rows with 'A'
| 0 | cloudy day |
| A | 1. info |
A | 2. more info |
| B | 1. info |
B | 2. more info |
|
|
2 Replies
Copilot isn't directly designed to extract tables from images. However, there are a few workarounds you can try:
1. Leverage Existing Tools:
Microsoft Excel Data From Image: While Copilot itself might not handle it, you can use Microsoft Excel's built-in functionality. Take a screenshot of the table and use the "Insert Data from Picture" feature in Excel. It attempts to convert the image data into a table format.
2. External Tools with Copilot Integration:
Limitations:
These approaches have limitations:
Accuracy: Extracting data from images, especially complex tables, can have accuracy issues.
Formatting: The extracted data might require further formatting to match your desired table structure within Copilot.I think a big problem is that Copilot wants to add more content. Sometimes it rephrases parts of the input. Or combine two bullet points into one.
I have not found a way to suppress this behaviour.
It seems that Copilot completely ignores negative prompts.
