I think screenshots will inevitably become a convenient record carrier，Whether it's in style saving or preview compatibility. The performance of OCR is no longer as tight as it used to be. Even most screenshot tools are easily capable of this effect. What logseq needs to do is to associate OCR-processed text with images. If these functions are introduced into the system, allowing content text retrieval of image assets will greatly improve the efficiency of extracting and organizing information.
When we put images with text into logseq, we can retrieve screenshot text when searching for keywords, even without manually deduplicating naming and tagging.
#Search image text by OCR scheme
open cradle
trim elm
You can use full text search with a simple query to find images using keywords in the name.
open cradle
You can use `full text search` with a simple query to find images using keywords...
What I mean by that is, we extract images through OCR, retrievable text, combined in some style into the property of the asset.
open cradle
Search image text by OCR scheme