waritkan/thai_handwritten_OCR
0
Thai Handwriting OCR
ระบบรู้จำลายมือเขียนภาษาไทยด้วยปัญญาประดิษฐ์ Thai Handwriting Recognition using TrOCR Deep Learning
Features
- Upload image of Thai handwriting
- AI-powered text recognition
- Based on TrOCR architecture with Thai SentencePiece tokenizer
Model
- Base Model: microsoft/trocr-base-handwritten
- Fine-tuned on: Thai handwriting dataset
- Tokenizer: Thai SentencePiece (30,000 vocab)
Development Team
Institution
Senior Project 2025 Data Science, Faculty of Science Chiang Mai University, Thailand
Tech Stack
- PyTorch
- Transformers (Hugging Face)
- Gradio
- SentencePiece
