Tesseract preprocessing -- Performance and Speed

[perito...@gmail.com]

Hey community,

i use tesseract for Text extraction but the i find it slow, so i have some questions to find out where can i contribute to make it faster :

- Did Tesseract process some Image traitement and preprocessing/cleanup at the start (need of Leptonica )? if it is the case what are those traitements? how much time do you think they consume ? and how could we disable them ?

- Is Tesseract convert all the image to tiff then process them ?

- Which part of Tesseract is the much time consuming ? and what are functions that you think we can remove or disable to make it faster ?

-  I find this article which propose some parallelisation in some functions to speed it up **[Performance Characterization and Parallelization of Tesseract Optical Character

Recognition on Multicore Architectures](https://pdfs.semanticscholar.org/dab1/23de2a9c25eaeaf7b6456116cea1e509f3f7.pdf)**, is it implemented ?

Thanks