Streamlit project with Tesseract OCR running on Streamlit Cloud.

1. Upload an image with text on it
2. Select the language
3. Select the image preprocessing options (if needed) and check the result in the preview
4. Crop the image to the text area (if needed)
5. Run the OCR and check the result in the text preview
6. Adjust the settings or image preprocessing and run the OCR again (if needed)
7. Download the result as a text file or copy from the text preview

Installed languages for Tesseract OCR

Streamlit application is working - 04.06.2024

• Change layout of the app
• Change checkboxes to toggle buttons
• Add cropping functionality: https://github.com/turner-anderson/streamlit-cropper
• Add more CSS styling
• Cleanup of python app and repository

• Use Pillow for image preprocessing instead of OpenCV
  • any advantages?
• Add Ace Editor for text preview
  • any advantages?
• Add other OCR engines and test them
• Add easyocr and test it
  • https://github.com/JaidedAI/EasyOCR
• Try tesserocr instead of pytesseract
  • https://github.com/sirfz/tesserocr
• Add PyMuPDF and test it
  • https://github.com/pymupdf/PyMuPDF
• Add ocrmypdf and test it
  • https://github.com/ocrmypdf/OCRmyPDF
• Add PaddleOCR and test it
  • https://github.com/PaddlePaddle/PaddleOCR
• Add keras-ocr and test it
  • https://github.com/faustomorales/keras-ocr

• Tesseract Documentation
  • https://tesseract-ocr.github.io/
• pytesseract Documentation
  • https://github.com/madmaze/pytesseract
• OCR with Tesseract
  • https://nanonets.com/blog/ocr-with-tesseract/

• https://github.com/JaidedAI/EasyOCR

• https://github.com/Belval/pdf2image

OpenCV is used for image preprocessing before running OCR with Tesseract.

• OpenCV Image Processing Documentation
  • https://docs.opencv.org/4.x/d2/d96/tutorial_py_table_of_contents_imgproc.html
• OpenCV Python Tutorial
  • https://www.geeksforgeeks.org/opencv-python-tutorial/
• OCR in Python Tutorials
  • https://www.youtube.com/watch?v=tQGgGY8mTP0&list=PL2VXyKi-KpYuTAZz__9KVl1jQz74bDG7i

• https://pillow.readthedocs.io/en/stable/

import cv2
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

# or
coefficients = [1,0,0] # Gives blue channel all the weight
# for standard gray conversion, coefficients = [0.114, 0.587, 0.299]
m = np.array(coefficients).reshape((1,3))
blue = cv2.transform(im, m)

• CLAHE (Contrast Limited Adaptive Histogram Equalization)
• https://www.tutorialspoint.com/how-to-change-the-contrast-and-brightness-of-an-image-using-opencv-in-python
• https://stackoverflow.com/questions/50474302/how-do-i-adjust-brightness-contrast-and-vibrance-with-opencv-python
• https://stackoverflow.com/questions/32609098/how-to-fast-change-image-brightness-with-python-opencv
• https://github.com/milahu/document-photo-auto-threshold
• https://stackoverflow.com/questions/56905592/automatic-contrast-and-brightness-adjustment-of-a-color-photo-of-a-sheet-of-pape
• https://stackoverflow.com/questions/39308030/how-do-i-increase-the-contrast-of-an-image-in-python-opencv
• https://stackoverflow.com/questions/63243202/how-to-auto-adjust-contrast-and-brightness-of-a-scanned-image-with-opencv-python

Methods to rotate an image with different libraries.

https://pillow.readthedocs.io/en/stable/reference/Image.html#PIL.Image.Image.rotate

from PIL import Image
with Image.open("hopper.jpg") as im:
    # Rotate the image by 60 degrees counter clockwise
    theta = 60
    white = (255,255,255)
    # Angle is in degrees counter clockwise
    im_rotated = im.rotate(angle=theta, resample=Image.Resampling.BICUBIC, expand=1, fillcolor=white)

destructive rotation, loses image data

import cv2
(h, w) = image.shape[:2]
center = (w // 2, h // 2)
M = cv2.getRotationMatrix2D(center, angle, 1)
rotated = cv2.warpAffine(image, M, (w, h))

non-destructive rotation, keeps image data

import imutils
rotate = imutils.rotate_bound(image, angle)

destructive or non-destructive rotation, can be chosen py parameter reshape

from scipy.ndimage import rotate as rotate_image
rotated_img1 = rotate_image(input, angle, reshape, mode, cval)