Text Detection with OpenCV in Python | OCR using Tesseract (2020)

In this video, we are going to learn how to detect text in images. We will learn how to detect individual characters and words and how to place bounding boxes around them. We will also look at how to detect only digits.
🚀🚀 My Urdu/Hindi AI KZbin Channel 🚀🚀
/ @murtazahassan01
Premium Courses:
✔️ Computer Vision Game Development Course:
bit.ly/3ttLZ2s
✔️ Computer Vision with Arduino Course:
bit.ly/3wzLB4m
✔️ Advanced Drone Programming Course:
bit.ly/3qs3v5g
✔️ Learn to Build Computer Vision Mobile Apps:
bit.ly/3uioY1J
✔️ Jetson Nano Premium Course:
bit.ly/3L8uIlF
⚙️⚙️⚙️--My Gear - ⚙️⚙️⚙️
👉 Complete Gear 💈: www.computervision.zone/tech-...
👉 My PC Specs 🖥️: www.computervision.zone/tech-...
👉 My Video Shooting Gear📽️ : www.computervision.zone/tech-...
👉 My Laptops 💻: www.computervision.zone/tech-...
👉 Educational Products🧑‍🎓: www.computervision.zone/tech-...
👉 KZbin Starter Kit 🔴: www.computervision.zone/tech-...
Follow Me:
TikTok: bit.ly/3Vo76OQ
Facebook Group: bit.ly/3irDcb7
Discord: bit.ly/3JvyxAM
Facebook Page: bit.ly/3IvpU7W
Instagram : bit.ly/3NdGME3
Website: bit.ly/3ICFTS0
Github: bit.ly/3woU6PS
#ComputerVision
#OpenCV
#CVZone

[@nithinprincejohn7762]

Your tutorials are awesome. I already shared your channel with my students. I visited so many channels for learning opencv and out of that your videos stands out.

[@murtazasworkshop]

I am glad i could be of help. and thanx for sharing

[@ishtiaqahmad4118]

By chance I entered last 5 years before in programming languages. You will note my background is Accounting. For now it is sufficient that I am 75 years old man but I am learning Django, Ai, Machine Learning and etc., Last five years I asked my instructor of Java (Note= At that time, I did not know A,B, C of any programming but I know a sufficient knowledge about Programming - I am a lot thankful of Respectable Great Zia Khan) that I can check cross medical bill of the any person, but he could not replay. I am very happy of this watching Video, I can solve this last 5 years dream can be solved by python with AI Insha-Allah. Now I am trying to launch my E - Commerce Website very soon with Python, I shall contact very soon with you for a big project Insha-Allah. Ishtiaq Ahmed Siddiqui KARACHI

[@WLeigh-pt6qs]

Hi Murtaza, I indeed learnt something new! It's amazing! I really appreciate your time and effort and sharing :) !

[@edSabio572]

Thanks Doctor Murtaza! This lesson is a class act!

[@Marek-Marek]

Great job! I was learn something new, thanks a lot.

[@akhilpandey]

Thank you, just what I needed.

[@Abcsciencemusic]

For windows , Instead of down manually downloading and installing just do : pip install tesseract

[@shuterino9454]

thx bro

[@jurajzovinec6897]

tx man

[@victorgiaff8625]

Tnx

[@jonascampos5564]

How can we know if that's official?

[@d.e.a.dgaming654]

Nah bro thats not official

[@surajprakash3558]

This is an amazing video and effort. Thanks a lot !

[@ADNANAHMED-eo5xx]

Amazing TUtorial man, thanks a lot

[@davidgarza7355]

This was sick!! Thanks, It was so easy to follow :)

[@aham3687]

Awesome video. I was looking for technology that could read image captcha and extract text in the backend. Then submit text to the server without users requiring to type the captcha and your video solved my issue.

[@choogiesaur]

amazing, would you share your code as an example?!

[@quasa0]

@@choogiesaur did you figure it out?

[@fakhruddintahery1561]

Great explanation. Thank you so much. That was useful

[@srvanikar]

wow....mighty helpful.. Thank you Murtaza

[@shailavijay1]

Thanks for your video. Very well explained :)

[@sohailali5741]

Hi Murtaza, Your videos are awesome and very detailed and easy to understand each and every part. Can you please make a tutorial on Semantic Segmentation with Keras and tensorflow?

[@murtazasworkshop]

i will look into it

[@sidhantipatil1930]

Great content! Keep going!😇

[@iaconst4.0]

MUCHAS GRACIAS MURTAZA!! ME AYUDASTE MUCHO!, TE DEJE LIKE!

[@johntamplin]

Good video. From 20:20 onwards you are using the variable 'x' for both the loop count and the x-dimension of the bounding box. You got lucky that it didn't impact the results of the code.

[@YazeedAlkosai]

thanks so much, Dear Murtaza...could you do a tutorial for detecting the shape feature and make the dimensions of the shape of the workpiece appear once the camera start to detect.

[@imadsaddik]

Thanks, this helped me.

[@gonzalohernan4169]

Thanks so much for this!

[@sudhakarp1]

Excellent explanation, Thanks Murtaza

[@animeshmaji9515]

Keep making video like this and we keep learning

[@1phong.djk.392]

wonderful. I comment before watching. Keep creating more wonderful tutor videos

[@murtazasworkshop]

Many thanks

[@mandrivaal]

You're a legend! thanks

[@DrWho2008t101]

Great Video!

[@vcjayan8206]

Great help..Thank you

[@harshitsrivastava9903]

Great work..very helpful!!

[@murtazasworkshop]

Glad to hear that!

[@makelabsindia3092]

your video and explaining method amazing ,keep it up

[@murtazasworkshop]

Thanks a lot

[@shreyanshdwivedi7871]

Hello Murtaza, It was good explanatory video which was easy to digest and understand, loved your work, I want to know that if you can make a video on creating the language model of sanskrit for tesseract from the certain bunch of images consisting of sanskrit text , that would be appreciated. Hope to hear from you soon.

[@shan2739]

Thank you very much for this video. Can you also make some video on Deep learning self driving car using raspberry pi and jeston nano

[@murtazasworkshop]

I have already started the raspberry pi series . once the basics are covered we will move to more advance topics like selfdriving, ML, DL

[@shan2739]

@@murtazasworkshop yea..I already got them waiting for you.. Thank you

[@jairajsahgal5062]

23:13 " I have no idea why they did this" made me chuckle.

[@akashrastogi6431]

great job and thank you so much

[@youcefahmanache3298]

Hi Murtaza, you see your loop to DISPLAY SOME SAMPLES IMAGES OF ALL THE CLASSES, if the folder names were categorical/strings instead of numerical, i.e. (sign1, sign2, sign3...etc instead of 0,1,2,3..etc) how would you iterate through these folders? would you still use iterrows? Thank you

[@mabo2031]

hi Murtaza I have question: I want to convert to pdf(with ocr text) only few page from multipage tif image - is it possible without any other conversion ? multipage tif - 70 pages - but I want to convert to readable pdf pages from 10 to 20? best regards :)

[@ashok_ign5623]

Thank you so much your all tutorials are helpful keep It up 😍😍

[@murtazasworkshop]

Glad you like them!

[@arthurswanson3285]

Very useful. Thanks.

[@SliderBlues]

Thanks for the tutorial! It was very helpful for me getting a handle on OCR and Opencv for a redaction project I was working on. Now looking at fictionalizing text in an image. Ideally would like to replace words, like for an address, with different ones using the same font and size for the text. Not finding any solid examples of this that work. Would love to hear your take on how that might be accomplished.

[@murtazasworkshop]

You could do it with opencv python but it you want an end product you might want to look at unity.

[@SliderBlues]

@@murtazasworkshop - Hi. Worked on it a bit and watched more of your videos but still stuck. Posted on stackoverflow and so far no responses that might work. If you have the time would like to know your high level ideas on resolution. Thanks! stackoverflow.com/questions/64541909/how-to-fictionalize-text-data-like-an-address-within-a-document-image

[@Jake79]

Very clear! TY

[@murtazasworkshop]

Glad it was helpful!

[@silviacolabianchi8261]

Hi! thanks for the video. What do you think of tesseract version 5.0.0. Which one do you suggest to download? Version 4 or 5? thank you

[@tecint]

Thank you!! Good Job !!!!

[@murtazasworkshop]

Glad it helped!

[@chickenz4604]

Nice video, btw could you give me suggestion when it comes to recognize handwritten digit and then manipulate the data?? should I use OpenCV & tesseract?? or tensorflow & keras? I am still not sure about this, since im new for this application. Thank You !

[@Mikasks]

So what did you use? I'm having this problem aswell.

[@Shraddhanath]

Thanks for this workshop. Could you also post a video on how to detect tables(ruled as well as not ruled) in document images or segregate tables, text, charts and figures. Tx.

[@murtazasworkshop]

Thanks. i will look into it

[@smrutidabhole961]

thank you so much sir...

[@WorkHardThinkBetter]

Hi Murtaza, Your videos are excellent. I need to ask where is 1.png since on your website I can't find it in the projects section. Thank you so much for everything!

[@birdeye700]

Excellent work! You might want to create functions to show digits, show alphabets and etc. This will improve readability and reusability of code.

[@andresnet1827]

Very good tutorial :)

[@murtazasworkshop]

Glad you think so!

[@shuaishuai2009]

good video~ Thanks for sharing

[@anajajkkkk8490]

Thanks sir!!

[@fixitunes30]

Awesome videos....Any plan for video on Deep Learning Chat bot in near future?

[@santibogo]

Hi murtaza! A question: If i only want to retrieve specific text from the image, how can i do it? i'm already have the coordinates of the text that i want to get

[@richmail]

take screen shot useing those cords, and send it via internet, tho, you need to know whee you store the images, hope this helped (srry for the bad spelling, I type fast)

[@rajeshmeesala9470]

Nice explanation

[@riztya7846]

that's so preety cool i maen tips with ctrl + left click, is so helpfull for me just learning python, thank you ! btw i'm already subscribe you 👍

[@kevinalexander4959]

used this with a webcam to take an image and parse text from (like a piece of paper or menu on a phone), then you can take the .png snapped by the webcam and analyze with image_to_string(), takes about half a second to read text presented and the webcam's settings has to be on point with brightness and all that stuff

[@mehmettatli105]

so we cannot use this one as a real time word detection metod? How many frames did you get when you use with your webcam? i just want to use it for real time detection ,if you answer to me. Thx a lot

[@hirdeshchhipa8238]

Pp

[@dimitheodoro]

The Binaries for Windows ( 1:06) has not this selection to download, so i cannot find how to download the file for Windows. Can you help me pls?

[@namtongdosa9626]

Thank you so much, its very helpful for me. Could you please make a tutorial about train model to detect text in identity image ?

[@murtazasworkshop]

I have one tutorial on classifying digits. you might want to have a look at that.

[@radoslavtotev4401]

Hey, thank you for the lesson but do you have it for linux because I want to use Raspberry pi 3 for similar project?

[@ambijat]

I want to use clipboard image for text read in Ubuntu Linux. Can you tell me how to do. Windows has other option of using pillow library, but that does not work in Linux.

[@MegaR97]

Hello, sorry I can't donwload the tesseract version you are using the page says "currently there is no official windows installer for newer versions

[@nikethadinesh7123]

Thank you sir for this video. Cud u also explain with handwritten text recognition?

[@colbrac720]

Thanks a lot !

[@innotech4079]

Hi, it's amazing but is there any train part? 1. detecting the text edges, 2. img to text training

[@Lindvni]

How do you guys fix the path problem? Everything looks fine ,until I run it. It just gives me an error

[@MrNadeem1984]

It doesn't work on text in larger and smaller font size. Can you help me how I can use extract those text from image.

[@goat6836]

Thanks Man

[@safabelhadjromdhane2922]

Hello, I would like to say thanks very much for this channel your videos have been very helpful while learning OpenCV and python. I'm having a problem while executing this code and I would like you to help me fix this "cv2.rectangle( img(x, y), (x+w, y+h), (0, 0, 255), 1) TypeError: 'numpy.ndarray' object is not callable." What do you suggest I do to handle it? Thanks in advance.

[@sanchimittal9614]

Amazing tutorial! Great content!! Can you do a video or suggest something for noise removal like unrequired non-text portion and other kind of preprocessing before text detection in images? Thank You

[@fortfanop]

I think denoising with autoencoders will do the job...

[@naresh2934]

thanks a lot.

[@9319964634]

Good video, can we also find micr font and store using this method?

[@MasterZeroFX]

Nice vid pahjeet

[@nicoarasamuel1173]

Love the video... I've try to make a project on wich I try to send a text to a monitor, more like a power point presentation or a teleprompter app... Do you have any suggestions...or can you make a video about something like that?

[@vittal255]

hello Sir, our videos are really informative. They motivate me to build things that I have in my brain. Thanks for those videos. I have small doubt can we implement the same project using raspberry pi instead of a laptop/pc. If yes please suggest the necessary modification I need to make. I am really to new to IoT and computer vision started learning about them

[@_kayos_]

did you find an answer?

[@raghavgarg1258]

Awesome, Please upload more videos...

[@murtazasworkshop]

Thank you, I will

[@aaryangupta4815]

Love ❤ this tutorial, can we do this same on live video also???

[@gawaderajesh]

Great...Murtaza's Workshop - Robotics and AI... Posted on Facebook....... :-)

[@EduardoAdameSalles]

I have a table of prices (with money values), and I have an algorithm in Python to calculate the final value (with the taxes I want for me, and rounded). But I write each value into the software, and then it returns me the value I want, and then I put in a excel table. Is there a way to take the values from the first table, run the algorithm and then return me a csv or xlsx table with the new values? You don't need to do a full search, I just want a way to go to learn (opencv, etc). In Python I just know tkinter, but the bases.

[@interpreter001]

may be possible using pythonautogui(controlling keyboard & mouse)

[@alooooshm]

Keep it up👍

[@ekizahmet]

Great video and great tutorial, Thank you! I wish I could see before. Btw Detection only Digits is not working. Does anyone have any idea why?

[@riyachougule6356]

That's really so useful ! Thanks a lot ! Can you please explain how we can extract the column of y point from "boxes"?

[@shashwatshankar6647]

Hello brother, first of all a great job with it. I need some help with text detection from a image. Actually i have to extract the headlines from a article present in a image form. But whenever a special character appears the data after it dosen't appear in a array. Can you help me?It would be great

[@dennisasamoah2213]

great again

[@firegames2741]

Thank you, Mortaza for such a useful video. I need help from you, can you convert captcha file to text. I'm trying, but not converting properly.

[@liberocabec47]

hello thanks you for your tutorial, i have a question, i try to make bounding box around sentence who have same b['top'] coordinate, i managed to make the separation to the word and the bounding box that correspond to it but i can't entend it to the sentence or to the same line of text, someone can help me ?

[@nisarggandhewar8242]

Its very helpful video, Thanks for preparing it, can you please prepare video on to detect text by designing & training our own model on some dataset just like digit detection, that will be more helpful.

[@murtazasworkshop]

Yes, soon

[@MadanLal-hl5mw]

@@murtazasworkshop l

[@tehreemsyed8621]

Hi, how can we extract each character and digit from the license plate?

[@adityanambidi5129]

Thanks a lot! Amazing tutorial, although I am getting one problem. it is not able to recognize bold text like a stop sing. So is there any way to detect such things?

[@murtazasworkshop]

For custom text you have to train your own model.

[@minecrafttipstricks9537]

Can i use this if i need to read a code from cells tru a camera? and then sort them by codes. is that possible?

[@aironium]

Is it possible to get the whole bounding box of a word by combining boxes from every character?

[@prashantrai5102]

thankyou sir

[@anmolpunetha3994]

There is no change when i add the config, it is still detecting text too. Can anyone pls tell why is it so?

[@avikro9235]

Windows binaries download is not available in tesseract documentation

[@ans_human]

can we use it for other languages? Like indic languages for example hindi.

[@agentkoko3988]

Would love to watch more ❤️ . . . I tried the same it worked but tesseract is not able to get small text or it detects it incorrectly. Actually the text wasn't that small it was a 1920x1080 screenshot and the font was of normal size but it only detects some of the folder/icon names very few and there were some wrong words as well, can we do something about it ?

[@vimsen2221]

I get following error. --- pytesseract.pytesseract.TesseractError: (1, 'Error opening data file C:\\Program Files\\Tessaract-OCR/eng.traineddata Please make sure the TESSDATA_PREFIX environment variable is set to your "tessdata" directory ---- Where do I set this?

[@choogiesaur]

I admit I havent watched the whole video yet, but I pored through the code examples - when I try to get bounding boxes with tesseract, although the character prediction is pretty accurate (tesseract guesses the right character), the bounding boxes that are returned are a bit shifted/innacurate. they are generally to the left of where the character actually is and sometimes includes a large part of the next character. yet in some cases, such as the letter "I" it seems to be detected perfectly. have you experienced this, know how to solve it?

[@MustajabAhmed-or8ud]

having the same problem, and I think it is a common problem in pytesseract.

[@MustajabAhmed-or8ud]

hello all, if anyone can answer. do we have to install opencv and pytesseract in pycharm when we have already installed using pip?

[@gowthamns8228]

Wow very good, but the problem is "If the text is very clear and crisp its is showing output correctly", But I want to know if the image has multiple data not only text for example "bills, taken calendar photo or any kind of images", How to print the string from that, I tried my self it's not printing anything, Any idea for this?

[@javierpachon4424]

Thanks for your tutorial, I want to capture an email from an image but Tesseract does not recognize the @ symbol, how can I solve it?

[@muthuji8053]

videos really helpful for me ,but if we apply this dull image or text are present in bottom that won't detected

[@sooryaram439]

Great video sir, I just have a small doubt. How can we detect text in live video stream using the webcam ?

[@livetop2749]

@Dharmendra Kumar Singh How much to deploy this app with camera detect?

[@lakshit6429]

same question