help needed for a similar usecase

#2
by yashvoladoddi37 - opened

im trying to work on the final dataset creation like your "kanji_meaning" dataset but im having some trouble trying to get the image and the english meaning in two columns perfectly. i have a jsonL file, each file having "file_name" and "text" . how to get it in a labeled way such that the actual image and the "text" appears in the dataset like how you have created in "kanji_meaning" dataset. please help me out?

Hey, have you looked at the GitHub repository accompanying this project?

The included notebook lists every step I took, you should be able to reproduce them: https://github.com/sylvainlapeyrade/english2kanji_diffusion/blob/main/english2kanji.ipynb

Let me know if you are still having issues.

Hey, have you looked at the GitHub repository accompanying this project?

The included notebook lists every step I took, you should be able to reproduce them: https://github.com/sylvainlapeyrade/english2kanji_diffusion/blob/main/english2kanji.ipynb

Let me know if you are still having issues.

I did look at it, tried the same thing, but still to no avail. I'm getting a parquet file which just shows the images without the english meaning with them. ive attached relevant photos, mind taking a look and helping me out please??

this is how my jsonL file looks like.
image.png

this is how parquet file looks like after executing the code you suggested
image.png

this is how im wishing for it to look like

image.png

this is how my code that you suggested looks like

image.png

note that in the above pic, my "kanji-jpg-path" directory contains the jpg files of kanji characters (without the stroke numbers) with the jsonL file at the end of the folder

please help me out with this. thank you

Solved here.

sylvainlapeyrade changed discussion status to closed

Sign up or log in to comment