# ReTest (리테스트)

> ReTest is an AI-powered educational exam paper workspace and error-review notebook platform. It extracts questions from exam papers (photos or PDFs), erases handwriting, and automatically generates high-resolution, printable A4 two-column test papers with smart QR code mobile grading.

## Core Value Proposition
- **Automated Problem Detection**: Multimodal AI vision automatically detects, crops, and sequences individual problems, mathematical formulas, tables, and diagrams without distorting aspect ratios.
- **On-Device Handwriting Removal**: Erases student handwriting, pens, pencils, and highlighter marks to restore fresh, clean problem worksheets.
- **A4 2-Column Standard Layout**: Formats extracted problems into standard CSAT/exam two-column vertical A4 printable PDFs.
- **Smart Mobile Scoring (Harness)**: Generates a QR code on each printed worksheet for students to submit answers on mobile and receive instant grading and explanations.
- **Cloud Synchronization & Multilingual**: Supports cross-platform continuity between mobile devices and laptop/desktop web workspaces across 50 languages.

## Primary Interfaces & URLs
- [Web Workspace](https://retest.ai.kr/): Main laptop/desktop editing, problem selection, and PDF generation workspace.
- [Mobile Web](https://retest.ai.kr/mobile): Mobile client for smartphone camera capture and touch-friendly reviews.
- [Copyright Complaints](https://retest.ai.kr/copyright-report.html): Official IP and copyright notice and takedown portal.
- [Full LLM Specification](https://retest.ai.kr/llms-full.txt): In-depth technical architecture, workflows, and FAQs for AI agents.

## Target Audience
- **Students & Test Takers**: Creating customized error-review notebooks (오답노트) for CSAT (수능), mock exams (모의고사), and school midterms/finals.
- **Teachers & Tutors**: Assembling supplementary worksheets, homework sets, and weekly test papers from existing materials.
- **Parents**: Organizing past exam papers and tracking student weak points.

## Technical Highlights
- **AI Vision Engine**: Google Gemini 2.5 Flash API with multi-modal layout analysis.
- **On-Device OCR & Processing**: Tesseract.js in-browser text reading and on-device neural handwriting removal.
- **PDF Engine**: Native vector PDF layout engine with precise typographic scaling.