hussaintamboli / python-image-to-text

Python program to recognize Text from Images using Google's tesseract-ocr

Installation

1. Install system dependencies

   $ sudo apt-get install tesseract-ocr

2. Download the latest Opencv from https://opencv.org/

3. Follow these instructions to install opencv http://stackoverflow.com/questions/15790501/why-cv2-so-missing-after-opencv-installed and set PYTHONPATH like it's said

   export PYTHONPATH=~/projects/opencv/release/lib:$PYTHONPATH

4. Install python dependencies

   $ pip install -r requirements.txt

Test

It will recognise text from test.jpg

$ python recognise.py

Contributing

python-image-to-text's Issues

TypeError: src is not a numpy array, neither a scalar

When I run the recognise.py, there happens a typeerror:
Traceback (most recent call last):
File "recognise.py", line 105, in
print get_captcha_text_from_captcha_image("./test.jpg")
File "recognise.py", line 55, in get_captcha_text_from_captcha_image
tif_file = preprocess_image_using_opencv(captcha_path)
File "recognise.py", line 87, in preprocess_image_using_opencv
bin_image_path = binarize_image_using_opencv(captcha_path)
File "recognise.py", line 78, in binarize_image_using_opencv
im_gray = cv2.cvtColor(captcha_path, 0)
TypeError: src is not a numpy array, neither a scalar

not working with large amount of data

hi,

i need to implement ocr for some financial docs this repo works well but only for few lines not for a complete table please tell me how could i change repo for below type documents

thanks

Need better accuracy

https://stackoverflow.com/questions/32473095/image-to-text-recognition-using-tesseract-ocr-is-better-when-image-is-preprocess