Re: [Gegl-developer] Contributing code that detect and change text/shapes in computer screenshot?



On Sat, Aug 31, 2019 at 03:11:19PM +0300, Gil Eli via gegl-developer-list wrote:
Hello,

I was previously contributed to solve gimp screenshot bug here:
https://www.gimp.org/news/2018/05/20/gimp-2-10-2-released/

I want to contribute again.

I am the developer of this product
https://www.youtube.com/watch?v=QaMUjMtIz_g

It include some algorithem to detect texts/shapes on the screen.
In order to work, it needs to scan the pixels and figure out what pixels
are background and not. then It can remove everything without the
texts/shapes, and it change the color of the texts to different color.

This looks like what conversion to djvu format would need.

-- hendrik


Now I was able to make use of nvidia cuda to improve the performance.
My new algorithem that use  nvidia cuda produce even better result. Here is
an example:

https://i.imagesup.co/images2/4cad91bb5ce384ab7a98715fca41140c415131b1.png

The efficiency is very high. It takes 5ms to produce the result.

It working without relying on any open source project/image processing
frameworks. Everything is done by me in low level including the idea how to
do it.

I am wondering how much it can be used for GIMP as part of a feature or as
a feature.

Do you need the code that can do such thing?

To be clear, I have the knowledge to:

   1. Detect any pixel that is text/shape of any language (no OCR needed)
   2. Change the color texts/shapes
   3. Remove the background around texts/shapes


I am wondering if this can be used as some tool for GIMP and if so then how
to get started.

_______________________________________________
gegl-developer-list mailing list
List address:    gegl-developer-list gnome org
List membership: https://mail.gnome.org/mailman/listinfo/gegl-developer-list




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]