leitro / digitsocr

Handwrite digits recognition using KNN algorithms with training real time.

digitsOCR

Handwrite digits recognition using KNN algorithms with training real time.

###Requirements:

• OpenCV 3.0
• Python 3

###How to use:

1. Start the program using command line "python3 ocrlearnKnn.py".
2. Write digits on a paper under the camera and check whether they are detected correctly.
3. Training your own hand writing data: first, press "x" button to pause the screen and the digits of its processing will show simultaneously; second, press "e" button and input the value of the digits with whitespace separated; then, press enter and the training data will be updated; finally, after many times of training, the program will detect your hand writing digits effectively.

###Watch my DEMO video: http://www.bilibili.com/video/av4904541/

digitsocr's Issues

image for example

AttributeError: 'module' object has no attribute 'ml'

Hi, when I run the code I meet this error:

File "ocrlearnKnn.py", line 6, in initKnn
knn = cv2.ml.KNearest_create()
AttributeError: 'module' object has no attribute 'ml'

Do you know how to solve it?

got error on while running on linux

VIDEOIO ERROR: V4L: index 1 is not correct!
Traceback (most recent call last):
File "ocrlearnKnn.py", line 68, in
frame = frame[:,:426]
TypeError: 'NoneType' object is not subscriptable

TypeError: 'NoneType' object is not subscriptable

TypeError: 'NoneType' object is not subscriptable

TypeError Traceback (most recent call last)
~/Desktop/.../ocrlearnKnn.py in ()
65 while True:
66 ret, frame = cap.read()
---> 67 frame = frame[:,:426]
68 rois, edges = findRoi(frame, 50)
69 digits = []
TypeError: 'NoneType' object is not subscriptable

run with jupyter notebook 5.50 in anaconda
python 3.6.5
cv　3.4.1
Macbook Air