Extract text from images with Tesseract OCR on Windows

[@josephc3080]

This is really good tutorial. I appreciate the care you took in going step by step, especially through altering the path.

[@GNS216]

This is the most helpful tutorial on Tesseract that I've found. Thank you.

[@TheJoinckim]

Very very good tutorial for tessseract for koreans and clear pronunciation. Thank you.

[@hkim644]

omg. I was watching your video to install Tesseract. Meanwhile, I was amazed that you can read Korean. I thought you chose a random non-english language to prove Tesseract works with different language. Amazed as a Korean. I am trying to learn how OCR works because I want to make an app that requires OCR. But I have no coding experience or anything even close to digital languages, I am having some difficulties. At least I was able to use Tesseract after watching this video. Thank you so much!

Thanks for this tutorial: I have had trouble with converting text in mayan language here in Guatemala, I followed your steps and voila! Next step for me is to figure out how to train a set of recognition for our local mayan alphabets. Thanks a lot.

[@iancardenas-spanishbutcomp4074]

Did you get to train it for a different alphabet? Can you help me? I'm trying to get OCR working for IPA characters recognition

[@TzKet4m]

Your voice makes me happy to browse youtube, so clear fuark

[@seung-wanson9447]

FYI, If we never add anything to PATH other than default one, it will not pup-up that edit selection box. So by looking your video, i need to manually make the entry by separating new one with ";" (semicolon) Afterwards, if i click the edit button, i get the same pop up edit box.

[@emmanuelvelasco8753]

keep making these videos man! interesting content

[@deepak223098]

Can you tell how to train our own dataset ??

[@R.t.a.s]

Thanks a lot for this but can i use this for manuscripts as well? And if so plz tell me how :)

[@philglanville3974]

Hi, a very good tutorial, but as mentioned by yourself, and a comment by another, ref batch folder/file processing , I can not see or find any uploaded tutorial video ?????

[@saikushalmandala6438]

thats a good video but, how to preprocess the input image and then pass through tesseract can u please help on it ASAP

[@ahmedfarouk8197]

you can change your pdf to a one tiff file instead of converting it to several png files

[@opheliafromlcf9509]

How did you turn each page of the pdf into pngs? Thank you for this high-quality video.

[@opheliafromlcf9509]

Alright, alright, I got that to work. Now I am wondering how you write the code to make it run all the pngs at once instead of having to do each one line by line, one at a time.

[@harmindersinghnijjar]

Hey there, you can use Snip & Sketch on Windows. I'm making a guide on just that currently.

[@pixelvader2451]

So, should I do it one by one? I have complete books, is there no way to do this for several images?

[@itsdannyftw]

What mic are you using? Great video, thanks!

[@rezkiy95]

Thanks for no bs tutorial!

[@davidpimental6704]

I need help with mixed language pdfs - English and Ancient Greek. Also, I would like to target positions within the image taken from a pdf file.

[@epochseven4197]

Interestingly enough, the default install path for the Windows x64 version is: C:\Users\username\AppData\Local\Programs\Tesseract-OCR

[@allirashna2072]

im kind of skeptical of allowing changes to hardware. is it completely safe?

[@danielveraec]

Thanks for the information. How can I install additional languages to the ones you sample? Maybe you already said it but my English is not very good and I didn't listen to it.

[@beastmonsterthing3]

thanks so much. easy to understand and so helpful. you're a legend

[@simunyugashakti5373]

Hi..Please guide me how I can retrieve the coordinate positions of the word that I retrieved from the image..

[@jarongaus]

Your instructions are phenomenal. You are amazing to explain computer commands and tricks. The only problem is that this program sucks and it is a nightmare to use it Its not your fault. Thanks so much for teaching so many tricks.

[@hyperventilate7318]

I have photographs of people with the date printed below, can this solution extract the date? I need to do this for 1000s of photos. (batch)

[@yllamaecataylo9282]

Can I actually use this to categorize a file into different folders? Btw, im using php so i dont know if it will work

[@a2zGodz]

How do u train the tesseract? Can u point me in the right direction with something I can use?

[@DFIRScience]

I'll try to do a video about that shortly. Until then you can check the documentation here: github.com/tesseract-ocr/tesseract/wiki/Training-Tesseract

[@prateekgupta2916]

Can u pls help me in training tesseract..,, for the sake of public help.. I will be very thankful to you

[@kevinsanti4091]

a video on tips on how to train tesseract would be great! anyway thanks a lot for this video so far! helpful for my first steps and really appreciated! I'm wondering if someone has already done -as something more looking like a sort of end user application rather than an in-the-field programmer use - (or eventually how to do it ) 1) an overlay of the pictured document and the ocr recognition in such a way that the original document remain displayed as it is but "highlight-able " or 2) aslo how to generate a parallel ocr document which keeps the letter positioning and layout in the space page of the ocr output like on the original picture and in case of a document keep the original cutted picture in case of difficulties and low confidence level in the recognition. like for example on graphs pictures drawings...

[@mrmikearmstrong]

Nice tutorial, makes everything nice and simple to handle - On another note, I want to call the tesseract.exe file from a .NET application that has just taken an image of some text, is there a way to get the output of the OCR as a string in the console? Or would I have to wait until the character recognition has completed, then go and read that text file at a later time?

[@DFIRScience]

Yeah, I'm pretty sure you have to read the file after. I'll check if you can output to pipe.

[@GermanPowershell]

Basicly nice Video. But why you open and use PowerSHELL ISE, and then don't use anything from Powershell?

[@KhalilYasser]

Thanks a lot. How can I add a new language after the installation?

[@luisguevara9292]

It helped me a lot. Thank you very much

[@knowsmynametoonobody9191]

nice video, it's what I'm looking for , So, thank you very much!😀

[@fabarchimilku4073]

Hi, how do link to the batch folder converting thingy?

[@etil2jz]

Really good tutorial, clear.

[@cohas3424]

제가 찾던 동영상이네요 고맙습니다. ^^

[@jennilthiyam980]

lstm_recognizer_->DeSerialize(&fp):Error:Assert failed:in file ../../../../ccmain/tessedit.cpp, line 193 i got the above error when try to perform tesseract.exe 3.jpeg ..\out1.txt -l ben plz help me out

[@gabrielbessa2575]

try completely uninstalling and dowloading a updated version :v hope it helps

[@jennilthiyam1261]

how to train the new language which is not in the language list

[@sebastienjurkowski]

Hi, we are looking for some knowledgeable with OCR, specifically for text from a Video feed. The text would appear most often distorted, non-horizontal and sometimes wrapped or partially wrapped. The text to be read is strictly a short sequence of number and/or letters. There can be multiple variations of those sequences in the same image. Contact me that rings your bell :)

[@aradsoltani4646]

thank you that was very helpful:-D

[@DFIRScience]

Glad it helped!

[@Barklo69]

what happen with the tutorial to make your own datatrainer :(

[@jennyf.2124]

Have you maybe tried out wether it also works with handwritten texts?

[@DFIRScience]

Hand-written text (block letters) will work, but not be very accurate. Ideally, Tesseract should be re-trained on whatever font you are focused on.

[@jennyf.2124]

@@DFIRScience I see, thank you very much!

[@atharvagupta9355]

hey, does anyone know how to scan multiple pictures in one go and measure the amount of time taken for the same? Thanks for the great video

[@sunnyraven4563]

can you please do the batch file video?

[@Bismillah_bismillah_bb]

i usually play trivia games and i want to use it there can u plz try to make a video on that?

[@prateekgupta2916]

Hi sir Much needed video.. Can u tell me how to train tesseract to identify specific font

[@venkateshdhande6318]

first how to create pdf to images

[@punnarajeev867]

can we convert captcha image into text

[@gabrielbessa2575]

Great tutorial! thx

[@mattchew2203]

How did you manage to get such fast results? It is taking me at least 15 seconds to OCR a full page...

[@DFIRScience]

The quality of your image will make a difference. Try around 300dpi. That will give you good recognition but should reduce processing time.

[@finestanime5878]

Thanks bro it is really helpful

[@DFIRScience]

Thanks a lot! I appreciate it.

[@aokaf]

please help me find how can i use it on MAC pleeeeease

[@mrcb1698]

Not sure if you will answer to this but i'd love if you could help me doing the powershell/batch code you spoke about at the end to make it work on a hole file. I'm currently trying but not success yet. Good video btw !

[@DFIRScience]

Hey there. Sure, I can help with that. I'll post back after recording.

[@iancardenas-spanishbutcomp4074]

@@DFIRScience did you make a tutorial for training the ocr to get another alphabet? I'm trying to get it to work with IPA

[@rodrigogutierrez7775]

can do this with a captcha image??????

[@danperryy]

What a great job.

[@mydulislam4218]

Thank you very much for your nice tutorial. Buy I would like to help with you that how to use this tesseract ocr without power she'll. How can I have can I use this very easy way that is either the first year I take the PNG or image then how to use is the tesseract another way so that I can easily without any complexity. After installation the it is a vector and the language platform how I can use this very easy way from the text and from the image.

[@sayankumardey6826]

Hi, please share this pdf file to download.

[@selvas7502]

how to convert multiple images from the folder. without giving image name one by one. is there is any commend to do it.?

[@harmindersinghnijjar]

Hey there, you can use Snip & Sketch on Windows. I'm making a guide on just that currently.

[@thesocialtalk1853]

hello, i want to use another language in tesseract

[@dipsikhaphukan5563]

Wanted this same thing using java ..Please help!!!!

[@mahmoodal-imam2892]

Thanks a lot, brother

[@AliMurtaza-hs2ct]

Warning. Invalid resolution 0 dpi. Using 70 instead and blank text comes. please help

[@DFIRScience]

What is your input file? JPEG? PNG?

[@AliMurtaza-hs2ct]

Png

[@DFIRScience]

You might try the solution here: stackoverflow.com/questions/42990139/tesseract-ocr-how-do-i-improve-result

[@AliMurtaza-hs2ct]

Thanks . It worked

[@sangjunlee391]

형님 감사합니다.

[@adoniskomplex91]

How can I increase the accuracy?

[@DFIRScience]

You will need to retrain the model based on your specific problem. I'm working on a video for training tesseract.

[@jaiksah]

the moment i type tesseract.exe --help, it opens the exe for installation ,don't know why

[@DFIRScience]

Try uninstalling, and downloading the installer from here: digi.bib.uni-mannheim.de/tesseract/tesseract-ocr-setup-3.05.01.exe

[@xlnyc77]

using powershell ? so its not really for windows? this is DOS. Did you ever make a powershell script?

[@cezarmoniz6579]

Congratulations on the video. I'm from Rio de Janeiro - Brazil. Great accent in English! Can we work with tesseract with PHP? By the way what's your name?

[@19perception83]

Excellent video, however, my output was dreadful. English, clear to see and it rendered about 90% fine, however, there are wingding style artefacts all over the place. A bit pants really. Can also render as different file formats with some more easily readable formatting (.odt) etc etc Will look for an alternative to compare against

[@DFIRScience]

If you'll be using the same types of input, you may want to train a new classifier on your specific dataset. For a random image 90% is not bad. I would make a filter script to clean the text and remove wingdings, etc.

[@randomvideosshideos8508]

but this is not detecting text from product images

[@DFIRScience]

Yes, there are a lot of situations where the current training will not work. You may need to create a training set based on the problems you are working on, and retrain tesseract with your problem set. I'm working on a video to make custom training sets for tesseract.

[@tobiaskarl4939]

also one has to set TESSDATA_PREFIX to "installdir\tessdata"

[@tkinter3160]

Sir ocr can extract text from video ?

[@gabrielbessa2575]

unfortunately no, but if you extract the frames and turn them into individual pictures, you can then execute the program and get the .txt files :3

[@hitachimonsta9553]

Thanks!

[@rachelludmir7169]

greet vidoe very clear . you have a vidoe on how to train tesseract ? please it can be very useful for me

[@nikhilgjog]

good info, but it would much better if the author could make a condensed video. He has repeated same info or provided unnecessary info at multiple places

[@adoniskomplex91]

I've used pdftoppm.exe from poppler. Works very well.

[@bj16162]

btw default windows ocr better than tesseract in my language

[@christianrazvan]

It doesn't appear that tesseract is any good

[@DFIRScience]

Default models are so-so. You'll definitely need to train on your specific problem. I've used default models for general ocr where high error wasn't a problem.

[@zardashtshwany3784]

tnx a lot

[@massivefins2597]

Tesseract is crud... Use Tabula and PDF's... You can select your tables also...

[@tasmia5243]

so it is easy to use to everyone and I am the one who is freaking out?!

[@송승협-b9g]

Korean?

[@silviotadeu607]

Wonderful Dad!!..lol

[@fabulusinvictus2198]

Suzy!!!!

[@mauroamorso]

tesseract 0001.jpg -l eng

[@proxy7362]

Tesseract OCR is terrible.

[@mohamedseddig5878]

how in all dir by one click