WHAT IS IT?
Receipt OCR is a Python OCR engine specialized in processing receipts. It combines two approaches: Tesseract for raw text extraction, and LLMs (OpenAI, Google Gemini, Groq) to turn a receipt image into usable structured data. The goal: go from a photo of a receipt to a clean JSON ready to be ingested.
WHY IS IT INTERESTING?
- Dual engine: Tesseract for raw text, an LLM module for structured extraction. Pick based on your needs and budget.
- Multi-provider LLM: Supports OpenAI, Google Gemini and Groq. No lock-in to a single AI vendor.
- Custom JSON schema: You define the extraction schema you want (merchant, date, line items, total…) and the LLM fills the structure.
- CLI + FastAPI: Usable from the command line for one-shot jobs, or through production-ready REST APIs for both modules.
- Dockerized: Containerized deployment is provided, making it easy to ship on the backend.
USE CASES
- Automate expense report entry from photos of receipts
- Extract purchase line items from receipts to push into an accounting tool
- Build a receipt extraction microservice behind a REST API
- Bulk-digitize a backlog of paper receipts into structured data