OMR Sheet

Detect contours using OpenCV to evaluate OMR Sheet

Working of OMRSheet.ipynb

1. Convert Image to grayscale

2. Remove Noise Using Gaussian Blur

3. Apply Thresholding and find out edges

4. Find out the OMR Sheet(Contour) in the image

5. Crop out the OMR Sheet part of image

6. Apply Thresholding and canny edge method on the cropped image

7. Now find out all the circular(specifically choosen shape of bubbles) contours

8. Sort them row wise top to bottom

9. Generate mask using Thresholded image

10. AND this mask with the Thresholded image and count non zero pixels

11. More Non Zero pixels means this Option is bubbled, so store it

12. Compare Bubbled answer with actual answer

13. Draw Contours on Correct options with red and green color(acc to filled bubble)

---------------------------------------------------------------------------

Working of OMRSheetRollNo.ipynb

Wroking will be same with few changes

1. This time we need to consider exactly 1 bubbled contour in each column(not in row)

2. To filter Circular contours more accuratly, shape matching is applied on each filtered contour to remove any noisy contour(if any)