Bangla OCR Correction

Project aim:

• Correcting OCR output on Bangla language

Workflow:

1. Creating Bangla Corpus
2. Pre-processing corpus, converting articles to lines
3. Convert text-lines to Images
4. Running OCR on these Images
5. Calculating Word Error Rate of OCR output using ground truth text-lines

Project tree

• seq2seq_OCR_Correction
• OCR
  • ReadMe.md
  • ocr_paralell.py
• Preprocessing
  • preprocessing.py
  • preprocessing_2.py
• Text2Images-Pillow
• README.md
• OCR_Evaluation
  • compare.py

Preparing Text Dataset

• Bangla Text corpus not availaible
• Wikipedia Article Scrapped for corpus preparation
• Find Scrapped Corpus here

Conversion Text2Images

Using pytesseract to convert images2text

Spell Checker for correcting noisy output for OCR

Relevant Link to blogs/repositories