ahmetozlu / signature_extractor Goto Github PK

Hi,
I don't understand why there are two constants when you do :
a4_small_size_outliar_constant = ((average / c1) * c2) + c3, why not a4_small_size_outliar_constant = ((average c) + c3 where c=c2/c1 ?

Thx

hey there! amazing code.
I have a question, where do you get those constant_parameter_1, 2, 3 and 4 values? ? I'm still trying to understand where those values came from.

Thanks in advance!

Excuse me,
After testing, I found that it didn't work well in many cases. Is there a better signature detection model?

Please, is difficult to develope a function that return true if a signature is present and false if not?

I need to understand if a document has been signed.

Thanks a lot

In StrokeFilter.py file function def strokeFilter(im, orient, freq, kx, ky): is giving error on following line

sigmax = 1 / unfreq[0] * kx sigmay = 1 / unfreq[0] * ky

Error Stack -

File "D:\OCR\signature_extractor-master\signature_verifier\verifier\__init__.py", line 12, in verify
    kp1, des1 = extractKeypoints(img1)
  File "D:\OCR\signature_extractor-master\signature_verifier\verifier\ExtractKeypoints.py", line 42, in extractKeypoints
    img = preprocessing(img)
  File "D:\OCR\signature_extractor-master\signature_verifier\verifier\Preprocessing.py", line 26, in preprocessing
    newim = strokeFilter(normim, orientim, freq, kx, ky)
  File "D:\OCR\signature_extractor-master\signature_verifier\verifier\StrokeFilter.py", line 21, in strokeFilter
    sigmax = 1 / unfreq[0] * kx
IndexError: index 0 is out of bounds for axis 0 with size 0

Hi.

Is it possible to retrieve the coordinates of the form (x,y,w,h)?? What??

Code identify table header and other shaped components. how to avoid those components?

Error: 'image' and 'label' must be the same shape
While using the label2rgb function. The labels have (x,y,3) dimensions. How is that possible?

I'm thinking to skip saving pre_version image and pass to the next step
So, I'm thinking to get a function to do this but not sure how to do this.
I tried to convert it as unit8 type but the colour is not correct.
I think it's realted to the types and normalization issues.
Could you advise how to get the function to do this? Thanks

img=function(pre_version) img = cv2.threshold(img, 0, 255, cv2.THRESH_BINARY_INV | cv2.THRESH_OTSU)[1] cv2.imwrite("output.png", img)