Extract Text from Video - images | Tesseract

[@jmarcelosr92]

Excellent video! thank you for sharing. I'm starting a little project and this came out very useful. I had a problem when executing the line [ text = pytesseract.image_to_string(my_example, lang="spa")] on the get_text function, I installed and reinstalled many times the pytesseract function thinking it whas a PATH issue. (as returned in my console. Turns out that when saving the image in the process part [name = "./image_frames/frame" + str(index) + ".png"] my compiler detected the [/f] of [/frame] as a special character and messed all up. in the end i just changed it for [name = "./image_frames/image" + str(index) + ".png"] in case it helps anyone trying testing this algorimthm

[@python360]

Ah thank you!, that could well help others - were you using Windows or *Nix by the way? 👍

[@monkey_see_monkey_do]

Wow! This is awesome! Now not sharing the source code in the descriptions below the videos no longer makes sense because THIS TUTORIAL!)

[@python360]

*CMK discovers Dr Pi and his evil plan to scrape code from videos* :o)

[@monkey_see_monkey_do]

@@python360 Well if it's about my videos then be sure you'll always get it for free)

[@hammadh2162]

I have a question as I came across your video. First, thank you for your content as it is very helpful. Is there another method to extract the text from a video without taking the snapshot or picture every 4 second?

[@python360]

Hello! Thank you for your kind comment. You could maybe adjust the timing to say 3 seconds, or 2 seconds or whatever works, but using this method there will need to be some sort of capturing of the picture in order to do the text recognition. Perhaps there will be an AI product that can do it as the video plays, but using Tesseract and Python, this is the only way I know of. If someone else reads this and can advise then that'd be neat!

[@hammadh2162]

@@python360 awesome, thank you very much for your feedback. Greatly appreciate it.z have a good one!

[@python360]

@@hammadh2162 No problem, cheers Hammad, good luck with your project!😄

[@rahulnema9896]

How to extract text, presentation, from Live online Videos, any extension, so that it's applicable for all online text extractor from video

[@python360]

Hi Rahul, that's quite advanced, check out this repo, you'll need Open CV and Tensorflow github.com/renukatamboli/text-extraction-from-video-frames

[@ffmax55]

Existe t'il un moyen simple d'extraire l'intégralité des textes inclus dans une vidéo ? (pas l'audio, ni les sous titres générés par KZbin) sur PC ou smartphone. Impossible de trouver cet info (pourtant ca se fait facilement avec une image et de l'OCR, et une vidéo ca n'est jamais qu'une succession d'images)

[@python360]

Oui. Avec KZbin API 👍

[@ffmax55]

@@python360 Merci beaucoup pour cette réponse, je vais essayer de me renseigner un peu plus sur KZbin API :)

[@python360]

Gonna chance, merci monsieur! 👍

[@python360]

@@ffmax55 kzbin.info/www/bejne/qJbWmXR6maZ0nrM

[@acb_gamez]

Couldn't you just use FFMPEG to get the frames from the video?

[@python360]

I want the text from the frames, not just the frames(images)

[@acb_gamez]

@@python360 I just mean to replace the first two functions you made. Then you use the get_text function.

[@python360]

I get you. Yes, you could do it that way - good shout 👍

[@learning_rust]

great intro!

[@python360]

Thanks for the visit

[@girlgotTECH]

i am trying to execute this on vs code can you tell me any of the extenions that are needed to be installed?

[@python360]

I use "coderunner" but it's not essential. VSCODE can get in a muddle with which version on Python to run, so always worth running your code from.the terminal first, 👍

[@michaelmody]

This is great. What about grabbing transcripts from u-tubes.

[@python360]

Yes, sure, I was thinking that might be a good topic so stay tuned.

[@zarg2289]

Hi I cannot install any of the libraries needed in this I get errors at every step I'm on windows, I have tried vscode, pycharm but no luck and cannot see what I'm doing wrong

[@ShreyaChavan-m9l]

change the path of your python download

[@saiyamjain0007]

Loved the tutorial

[@python360]

Thank you. Much appreciated 👍👍

[@mplinterviewhighlights4103]

Hi Bro, good video! i was wondering if we can just display the text where the frames between 1 and 2 (Both same) only display once instead of double? So the pattern of the frames output is (1,2,.......2,3.......2,1......) instead of (1,1,2)

[@python360]

Yes, so if the frames haven't changed, then it doesn't give a duplicate? You could parse the text and if it's identical to previous then you would not save it, or just measure the similarity between the images before parsing: eg. www.geeksforgeeks.org/measure-similarity-between-images-using-python-opencv/

[@mplinterviewhighlights4103]

@@python360 Thanks for the advice! Yes, i want the option to have the text to display between frames and not continuously, if i remove the duplicate image then that image result will not ever be printed out in the output. so if text A appear first, it wont appear next as text B will take place and after text B, text A have the option to be printed out or not

[@kivachervi443]

Hi there! I am brand new to python, so this may be a stupid question, but how do I install libtesseract-dev? I was able to install tesseract using pip, but cant figure out how to do libtesseract. I am on a Mac, by the way. Thank you!

[@python360]

Hello, try : $ brew install --with-libtiff --with-openjpeg --with-giflib leptonica $ brew install --devel --all-languages tesseract I don't know so much about Mac, but also maybe you can do : brew install --libtesseract ? (You're right though, It won't be pip, as it's an actual system library that you need)

[@kivachervi443]

@@python360 Thank you, I will try this out and see if it works. On a windows PC, would it be any different?

[@python360]

No. Just change the forward slashes in the filepaths to suit your particular Windows paths, you can use VS code (get relative path - right click over file ) if you have it/use it.

[@kivachervi443]

@@python360 Brilliant, I really appreciate the fast response. I'll let you know how it goes.

[@harshavardhanranger]

Hey @@kivachervi443 , how did it go? were you able to make it work on mac ?

[@naumanasif383]

Great. If I want to extract urdu text what should I do then?

[@python360]

I believe you can specify different languages, might need to check through documentation and see if Urdu is listed.

[@sandeshchand8729]

Hello python 360, I am facing one problem when I am trying to divide video into frames the text in it is replicating in frames I am getting the same text extracting again and again ".Is there any best algorithm or method to divide video to frames effectively.

[@python360]

You'll need to know how many frames per second and then use the modulus operator, or...Try some sort of image detection with openCV to only extract text when the image (frame) is noticeably different to the previous. With my project I worked out the number of seconds between each different page in the video, so it was constant which admittedly was much easier.

[@sandeshchand8729]

Could you explain more about modular operation.Thanks

[@python360]

@@sandeshchand8729 Yes, so a = 13 b = 5 # To store the remainder obtained # when dividing a by b, in c c = a % b c = 3

[@thenextbigthing1393]

is it possible to use this to extract text from a Movie?

[@python360]

It depends on how clear the text is. Subtitles of the film would be ok. You'd just need to get the timing right and check you haven't extracted the same text too many times 👍

[@rezacute1]

tanks for video . I want to write python app for read Instagram story insight reach from recorded video can u help me to write this app?

[@zunairakhalid7358]

Can anybody help me if i want to convert the text from video as an audio output?

[@python360]

Most of the TTS stuff you have to pay for from what I can see. Have you tried pip install gTTS ?

[@vishnuk5680]

sir can you make a code that can extract text from video frame by frame and then convert that text to subtitles with time stamps can you do it, its a kind request 🙏 i am unable to this bcoz i am a beginner

[@python360]

You can get the transcripts from the KZbin video itself, without any need for code. (Assuming you're on a computer rather than phone?).

[@gamingsirver]

Now if only I could extract the code text from this video so that I could use it to extract text from videos.

[@python360]

Email me I can send it to you 👍😉

[@girlgotTECH]

@@python360 can you send me that as well?

[@benyaminma8564]

I love this .🍻

[@python360]

Thanks man!

[@interviewsyllabus1111]

HEY ! Can i get the code

[@python360]

github.com/RGGH/OCR/blob/main/videx.py

[@rushikeshsherekar9850]

for a minute i thought why is jhon stamos teaching python .... ps: ik i spelled it wrong

[@python360]

Lol! 🤣

[@lordmelbury7174]

Tremendous! I have some videos I need text extracting from, I'll send you an email!