Table Recognition System
Deep learning system for borderless table detection and recognition in PDF documents.
- Role: Researcher
- Timeline: January 2023 - May 2024 (M.Tech Thesis)
- Team: Under Prof. Guidance at IIT Delhi
- Technologies: Deep Learning, Python, MTL-Tabnet, PyTorch
- Link: https://github.com/Laxman824/MTP2-work
Problem
Tables in PDF documents, especially borderless tables, are difficult for screen readers to interpret. Existing table detection works only for bordered tables, leaving visually impaired users unable to access tabular data.
Solution
Developed a deep learning model (MTL-TabNet architecture) that detects table regions, identifies rows and columns even without visible borders, and extracts structured data for accessibility tools.
Impact
- 85% accuracy on borderless tables
- Published research at IIT Delhi
- Integrated into accessibility pipeline
- Open-source contribution
Key features
- Borderless table detection
- Row and column segmentation
- Cell content extraction
- Multi-page table handling
- Export to CSV/Excel
- Integration with screen readers
- Scanned document support
- Table structure reconstruction
Tech stack
- Ml: PyTorch, MTL-TabNet, Transformers
- Cv: OpenCV, Detectron2
- Data: pandas, NumPy, scikit-learn
- Tools: Weights & Biases, Jupyter, LaTeX
What K Laxman learned
- Training custom object detection models
- Document layout analysis techniques
- Academic research methodology
- Accessibility in AI applications
Explore more
- Home — overview, skills and a built-in AI assistant
- Experience — roles at Think360 AI (CAMS), CAMS Mutual Funds and IIT Delhi
- Projects — GenAI, LLM, RAG and full-stack builds
- Education — IIT Delhi, M.Tech & B.Tech Computer Science
- GitHub Activity — open-source contributions
- Contact / Hire me