Creating Donut Dataset

Hi, I am trying to generate a donut docvqa training set, however I could not create the DatasetDict object.
I have a Pandas DataFrame that has a column for images, which have the PIL image, one for query (string) and one for the answers (string).

I tried the Dataset.from_pandas, but it did not work.

How can I convert my training data to DatasetDict?

Just figured out that with the use of Dataset.from_dict() method, I can solve my problem. Replying my own question in case somebody else has the same problem in the future.