Image Features and Image Alignment | Getting started with OpenCV series

[@sanjay4675]

If we want to do image classification for multiple invoice types is it the way we should proceed ?

[@christianrazvan]

But if you don't have any images on the document those features comparer fail miserable, for 2 similar reports for instance

[@PRAKASH77781]

Thanks for the video. However, found this error with matches.sort(key=lambda x: x distance, reverse= False) command. It is returning"tuple has no sort algorithm". Kindly share your throughts

[@ronaldkclai]

Hello, same error here. Used sorted() and same. Any hint please?

[@alvaroretana1243]

Same Error! -> AttributeError: 'tuple' object has no attribute 'sort'

[@FancyVolcano]

Looks like the match() function returns a tuple. Converting it to a list would do the trick. matches = list(matcher.match(descriptors1, descriptors2, None))

[@chriss6334]

The way Liben has described it should work. But to add another solution to the pile I'll chuck in what I did. I replaced the "matches.sort(...) line" with this: matches = sorted(matches,key=lambda x:x.distance, reverse=False)

[@NutsandGuts]

matches.sort doesn't work with tuples in python

[@ilovebyungsoo]

Try this instead matches = sorted(matches, key=lambda x: x.distance)

[@BCTAHbKA]

Das ist das geilste Video ever

[@tapudatta5615]

This doesn't seems to be a efficient way .How to finetune the parameters for different images

[@LearnOpenCV]

The approach shown in this video is done using classical CV - i.e using opencv. There are different methods available using deep learning models too for image alignment.

[@santhoshreddy4789]

can you please give source code

[@MATLAB1Expert1]

hey the source code is availabale in udemy course

[@sanjay4675]

@@MATLAB1Expert1 Is it the same content or something extra is there in course ???

[@rutujanaik2321]

hello can you send 13 and 14 th line code again

[@timmarkhuff]

I am still working on this, so I am not sure, but I believe they are just: im1_display = cv2.drawKeypoints(im1, keypoints1, outImage=np.array([]), color=(255, 0, 0), flags=cv2.DRAW_MATCHES_FLAGS_DEFAULT) im2_display = cv2.drawKeypoints(im2, keypoints2, outImage=np.array([]), color=(255, 0, 0), flags=cv2.DRAW_MATCHES_FLAGS_DEFAULT)

[@FancyVolcano]

@@timmarkhuff looks like it is "flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS" im1_display = cv2.drawKeypoints(im1, keypoints1, outImage=np.array([]), color =(255, 0, 0), flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS) im2_display = cv2.drawKeypoints(im2, keypoints1, outImage=np.array([]), color =(255, 0, 0), flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)

[@rutujanaik2321]

DRAW_MATCHES