OPTICAL MARK RECOGNITION (OMR) MCQ Automated Grading- OpenCV Python
ฝัง
- เผยแพร่เมื่อ 13 ก.ย. 2024
- In this video, we are going to learn how to create an Optical Mark recognition algorithm in python using OpenCV. We will write the code from scratch going step by step while discussing the details of each line. We will use the webcam to automatically find the grades of MCQs.
🚀🚀 My Urdu/Hindi AI TH-cam Channel 🚀🚀
/ @murtazahassan01
Code and Files:
coming soon...
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.computervi...
👉 My PC Specs 🖥️: www.computervi...
👉 My Video Shooting Gear📽️ : www.computervi...
👉 My Laptops 💻: www.computervi...
👉 Educational Products🧑🎓: www.computervi...
👉 TH-cam Starter Kit 🔴: www.computervi...
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
You offer all these courses for free, you are indeed a great Man. Thanks for the awesome work Sir.
I've been watching your tutorials since I woke up today and I'm really amazed!!! Tks for that!!!
Glad you like them!
This happened to me today 😂
Honestly, you have helped me SO much with your videos! It's difficult to find good documentation for these things and the way you break down the details is so amazing. Thank you, you've taught me a LOT! I'm for sure going to get some courses from you.
Thank you for your incredible tutorials! Your selfless dedication to sharing knowledge for free is truly appreciated. You've made a positive impact in my learning journey. Grateful beyond words!
This video is one of the best I have ever seen here on TH-cam! Your explanations have been excellent, I learned many things. Now it’s my turn and I try to do that myself. Thank you very much!
Please tell me the code he is teaching is actually working or not ? If its working only then i will make this project.
Great Tutorial. I wonder why this video has so few likes and comments. This video and the creator's effort need more appreciation.
Thanks Murtaza
I'm hitting the like button multiple times. Please keep uploading projects like this sir :)
I admire the punctuality of your course. Long ago (1985) our School Guidance Bureau bought an OMR reader and the Sofware to read bundles of marked sheets, marked by students, to establish their level of performance on several multiple choice tests. The raw scores were saved to a file. And by another program the final results were calculated and printed.
I wonder if the same can be accomplished by the scanner in an all-in-one printer :)
Until now the sheets I made myself worked splendidly with your program :) THANKS!!! :)
your explanation and concept are excellent. I learn a lot from you.
but why are we using all that sorting function to sort the 4 corner points when we can easily get it using boundingRect()
here is my code, hope it helps!
for c in contours:
if cv2.contourArea(c) >10000:
peri = cv2.arcLength(c, True)
approx = cv2.approxPolyDP(c, 0.02*peri, True)
x, y, w, h = cv2.boundingRect(approx)
# this will help to get the 4 corner points.
_1st = np.array([x, y])
_2nd = np.array([x+w, y])
_3rd = np.array([x, y+h])
_4th = np.array([w, h])
############ SORTING IS DONE ##########################
But if you want to check if the 4 points are landscape or portrait and want to change in the landscape. then use this code after getting the 4 corner points.
def findlen(p1, p2):
"""
using Pythagoras theorem to get the length of the two points
length = sqroot[(x2-x1)**2 + (y2-y1)**2]
"""
lenth = ((p2[0] - p1[0])**2 + (p2[1] - p1[1])**2)**0.5
return lenth
paper_length = findlen(_1st, _3rd)
paper_breadth = findlen(_1st, _2nd)
if paper_breadth > paper_length: # here breadth should be smaller than length
temp = _1st.copy()
_1st = _3rd.copy()
_3rd = _4th.copy()
_4th = _2nd.copy()
_2nd = temp
############ DONE ##########################
Also, what if the image is flipped (upside down)
to fix that use some reference on the image and can easily fix it.
If you have a better answer and if I m wrong somewhere I m happy to learn from you. Thanks a lot.
Please tell me the code he is teaching is actually working or not ? If its working only then i will make this project.
Honestly, you have helped me SO much, Your explanations have been excellent, Sir how can we detect none marked bubbles and multiple marked bubbles?
Excellent presentation. Thank you for sharing. I will be watching out for all of your upcoming videos.
What an project bhai, superb work. Much much impressed. Thankyou for sharing ur knowledge and approach towards the opencv.
Thanks for sharing your experiences with us.
Great video! Everything is perfect!
Você é inacreditável, sou seu fã
Great project tutorials.Please upload more projects. U are best man .
I think a new hero has born.
it's very helpful and clear!! Thank you!
You are doing well , your effort and clear explanation is really need to appreciate 👏 🙌 👍 😊 Thumps Up for your great work 👍 😀 💪
Great video! Is there a way to identify a corrected answer? i.e. 2 answers are marked but 1 is with an X on it, I only want to identify the one without the X on it....
hank you soo much very direct link n works for me love the way you expressed the installation .
Nice tutorial. Keep on uploading great project.
Thanx. Appreciate your support
Many thanks for this amazing video
Awesome explanation, much helpful
You are awesome ❣️
Great Tutorial keep uploading this type of content ... please
Wow! it's Amazing I did it well ! Perfect work !
I could not stop clapping ... I needed this so bad. but my question is what if none of the options were marked by the student?
You can set a min threshold and if its not met then the result is zero.
@@murtazasworkshop Thank you..exactly how I was thinking. thank you very much. i am building this in python Flet and you have done the most difficult part for me.
perfect Video, I really big fan of your channel
Excellent Video. Keep it up man
Thank you
MasyaAllah, terimakasih banyak bro, tutorial yang luarbiasa🤩🤩🤩🤩🤩🤩.
you are very very intelligent ... Thank you so much
Really helpfull tutorials. Helped a lot
Sir Murtaza, thank you for the great information. I need an OMR scanner to save the database as CSV files and to work with the MySQL database to save the results.♥
loved it.thank you so much!!!!
Hello Sir, I’m using my own OMR that is a bit different from yours.I did up to finding threshold image and got a really nice clear picture. However, when I try to find the splitting of thresh image. I’m getting a really weird image.
def splitBoxes(img):
rows = np.vsplit(img,5)
cv2.imshow("Split",rows[1])
utilis.spliBoxes(imgThresh)
The above line of code gives some portion of row 0 image and some portion of row 1 image in rows[1]. Why is it not splitting each row correctly and what am I doing wrong.
Hoping if you could provide some guidance.
I took two months to learn Adobe After Effects and Premier Pro. Now I am learning soft soft to make soft so I can have more verity
if anyone is having problem splitting the image horizontally or vertically try resizing imgThresh to the nearest multiple of 5(or another number depending on your needs)
def splitBoxes(img,choices,questions):
imgWidth = img.shape[1]
imgHeight = img.shape[0]
multipleW = choices * round(imgWidth / choices)
multipleH = questions * round(imgHeight / questions)
img = cv2.resize(img, (multipleW, multipleH))
rows = np.vsplit(img,20)
boxes = []
for r in rows:
cols = np.hsplit(r,5)
for box in cols:
boxes.append(box)
return boxes
this is how I did it
khouya , wakha duz prive layhfdeek , whatsap wla telegram plz
This so much help for me
Thanks a lot 🤟👍
@@yuriarfil5154 you're welcome
Hlw bro ...if I want to increase number of questions then how can I do can you explain pls
Hi Murtaza, thanks for your video - it was very informative! I have a question - if you instead show the whole A4 paper in the webcam, then would the code break? This is because the 'biggest contour' will be the A4 paper itself, and the second biggest contour will be the questions box. How can you adapt the code to constantly work regardless of whether the A4 sheet is fully in view?
I love this time for the enlightenment
Nice tutorial! can you tell me how to detect square boxes in an image?
thank you for great tutorial sir, can i use this code for multiple point, in this tutorial have two point biggespoint and gradepoint. can i use this code for multiple biggespoint, example there are 4 biggespoint in same weight and height
Great work! Thanks
Thank You, Sir, A world of thanks.
You are the best! Thanks a lot!
That was amazing 💙
great video.
I appreciate if you increase IDE font size.
In The setup yup that was my problem thanks a lot
Sir,
Instead of using options in ans variable .we have any chance to compare two images for correction.
return _nx.concatenate(arrs, 1)
File "", line 6, in concatenate
ValueError: all the input array dimensions for the concatenation axis must match exactly, but along dimension 0, the array at index 0 has size 255 and the array at index 2 has size 350
Hi! My name is Flávio, i like so much your tutorials. I have a question, how can i detect forms like vaiations graphs? For example, variation graphs like heart beats
Thanks Sir! Is it also applicable with radio buttons? Because I want to get the value of the selected radio button on the pdf file
hi sir important question can i turn this code in to an app if eever i made my onw version? and add some things?
Much appreciated and Thank you for sharing (In subscribed as well)...however I have a question, this Pycharm requires me to pay, are there any other Free alternative python I.D.E's to create these great projects?
hello im new to machine learning and i have a project now in my school: execute sql code from image, what can you suggest me to learn first
Dear Sir! How to edit the code to have 4 answer choices and more than 5 questions
i need to get rectangle covering the whole OMR sheet by 4 black circles on the 4 corners od the sheet. How can i do this?
can u help me sir i need the code for utilis or else if anyone find it can u share the link please guys
thank you so much , it worked 🙂
Thanks so much for tNice tutorials bro
If any dot or spot remains in the gap part of an OMR sheet not in the number and circular part of OMR, Will it be a problem or will the machine count it?
i appreciate that. thank u!!!
Nice Tutorial. Can you do a 100 items OMR tutorial?
you can expand this to as many mcqs as you require.
The reason for my asking, is due to crackling/distorting soft, whenever the drum/snare, or whatever is playing. TNice tutorials was the case off
Bro you haven't uploaded code.
The OMR code is running fine 5x5 Matrix and if all rows bubbles are fill. I am facing problem when I try to run 6x4 or others and few rows bubbles are empty.
Please suggest
Sir what if the question paper contains multiple columns of questions can please suggest me a solution and one more thing is that this code only works best for question paper containing equal number of questions and choices
sir if we want to increase no. of question ?what are the change that we made?
File "", line 6, in concatenate
ValueError: all the input array dimensions for the concatenation axis must match exactly, but along dimension 0, the array at index 0 has size 255 and the array at index 2 has size 350
wow such great learning
Gald i could help
Thank you very much
Thank you sir very helpful tutorial you provided to us. Sir, I have one question-How do I extract answers, if image is rotated by any angle?
If its small angle this could work. But if the angle is big then other methods would be required.
@@murtazasworkshop Thank you sir for reply. Can you give a hint to handle a big angle?
Where can i get the source code?
the top, when I installed soft soft (restart didn't help). I have a creative softblaster z softcard. I'm assuming it has sotNice tutorialng to do
Which application are u using pycharm ???
Sir how we read and store this data into database or any file kindly explain it
Hi, thanks for such a great video, please let me know can it read the barcode too?
yes you can read bar-code as well but you will need a separate package for that such as Zbar
Excellent
If bubble (mcq filling place) has been cut with blade then the remaining mcqs will scan omr machine
Reply me
Sir please for the stackimages function, can I get the code fro line 37 and 38
Hello After i installed utils I get the error message : module 'utils' has no attribute 'stackImages' what am I doing wrong?
awesome!
Thank you so much, sir.
You are most welcome
Honestly it help but how we save the result in excel and make it a list and also it can take name of the candidate and there roll number and save all these in excel.
Hello sir, I have faced a serious problem, that I want to implement dynamic rows. How I can accomplish that.
Thank You.
Horizontal and vertical split not working properly. They are not giving one row?? What should i do
I was surprised how understandable tNice tutorials tutorial is, thanks!
it cant work dude
cv2.error: OpenCV(4.2.0) C:\projects\opencv-python\opencv\modules\imgproc\src
esize.cpp:4045: error: (-215:Assertion failed) !ssize.empty()
why??
it seems your image path is wrong.
could you solve it?
@@daparthiharshavardhan2748 change cameraNo = 1 to cameraNo = 0
@@istiaqahmedjoy1014 change cameraNo = 1 to cameraNo = 0
@@yudhistiramuslim3117 hi, i don't see camerano, how to? hhuhu
Sir i want to make this project on flutter so sir is this possible in it .. plzz replay .
Hello I encounter an error in img= cv2.resize(img, (widthImg, heightImg)) how can I resolve it? Thank youu
Thanks for this video, but how can I expand this to many mcqs?
same question. have you figured it out? I got 60 items since it is a standardized test.
Aoa sir kindly tell me what model is used in this project
Thank you so much!!!!
You're welcome!
thanks for your great tutorial
may i have the source code? im a student and currenty working on a project for training
Is 100 mcq tutorials are stress full.
How to get the utils / stackimages function?
Can I ask you a question.
approx = approxPolyDP (i, 0.02 * peri, True)
python does not recognize approxPolyDP as a function.
It only accepts approx. I use python 3.7.2
use cv2
approx = cv2.approxPolyDP(cnt, 0.02 * peri, True)
Hi, is this neural network?
sir how can i implement that on mobile camera?
im building smart bubble sheet mobila app
great, but if there's any question that students don't fill in, we will be at fault because we are looking for the maximum value of a row
this is just an example code so all conditions are not considered. This condition that you have mentioned can be easily taken care of with a few lines of code. Where the system would check for a minimum number of pixels to count as marked.
On 1:11:43 ...
Try something like this:
if np.amax(arr) >= 5000:
myIndex.append(myIndexVal[0][0])
else:
myIndex.append(-1)
coooool!!!