Templates
Handwriting OCR Pipeline
Deskew, binarise, segment, recognise, correct.
A transcription pipeline for real-world documents rather than clean scans: automatic deskewing, adaptive binarisation that survives uneven lighting, line segmentation, per-line recognition, and lexicon-based correction with character error rate reporting so you know whether the output is worth reviewing.
What is inside
- Automatic deskewing before segmentation, which is where most pipelines quietly fail
- Adaptive binarisation and denoising for photographs rather than scans
- Ruled-line removal via morphological opening, preserving letter strokes
- Projection-based line segmentation with per-line recognition
- Domain lexicon correction that only auto-fixes unambiguous matches
- Character error rate reporting per page, so quality is measurable
Built with
- Python
- OpenCV
- PyTorch
- Transformers
- FastAPI
Next step
Have an AI problem worth solving?
Tell me what you are trying to build. I will tell you honestly whether AI is the right tool, what it would take, and how I would approach it.
Typical reply within 24 hours
Working from Dhaka, Bangladesh · remote worldwide