Back to Projects
Safety Scanner is a tool that digitizes handwritten safety documents using machine learning. It uses a HuggingFace TrOCR (Transformer-based Optical Character Recognition) model served through a Python API to convert handwritten forms into searchable, structured digital text.
The system handles real-world handwriting with all its inconsistencies, processes images through a preprocessing pipeline to improve recognition accuracy, and exposes the results through a clean REST API so other systems can integrate with it.
Key Features
- Handwritten text recognition using a transformer-based OCR model
- Image preprocessing pipeline to improve recognition accuracy
- REST API for easy integration with other systems
- Batch processing support for multiple documents
- Structured output for downstream data processing