doctr

Experiment and integrate with different OCR frameworks seamlessly

A OCR labeling tool - made for docTR

Spacy, HAC, pytesseract, easyocr, doctr, mmocr, layoutlm, paddleocr

Simple Streamlit application that parses the data from Invoice images and returns it in JSON format

Cookiecutter template for Python 3 packages

Modular OCR pipeline for live-camera documents/screens. Auto-aligns perspective, blocks obstructions, and extracts text using interchangeable deep-learning models.

A local-first engine for turning scanned textbooks into structured RAG databases. Features layout-aware OCR, deterministic math cleanup, and pedagogical role-indexing using LanceDB and CLIP vision.

Микросервис, позволяющий распознавать информацию со справок, учитывая в том числе табличное форматирование

Analysis of funding under the NHS Canada

How OCR errors break biomedical NER: BC5CDR abstracts rendered as degraded scans, read with docTR, gold spans projected by character alignment and tagged by PubMedBERT; entity F1 vs CER with bootstrap CIs and three fixes (OpenCV clean-up, noise-augmented fine-tuning, local-LLM correction).

Arabic OCR pipeline with fine-tuned ONNX models for detection, recognition, and orientation correction. Runs on GPU via ONNX Runtime, with optional LLM postprocessing for low-confidence words.

A cross-platform docTR OCR utility for .NET using Python.

Production-grade OCR evaluation and optimization platform for benchmarking PaddleOCR, DocTR, and Tesseract with preprocessing optimization, statistical comparison, experiment tracking, and interactive HTML reports.

Drop in a resume PDF, even a scan, and watch OCR + an LLM turn it into structured JSON on one GPU, deployed with chalkcompute

Admin Page Ecommerce helps understand the login flow and table management. It follows the MVC OOP model and integrates libraries like doctrine/dbal, bramus/router, bladeone, validation, and phpdotenv.

Document AI for invoices, receipts and bills.