emileegraphic698/VisionFusion_OCR_QR
🧠Unlock data insights with VisionFusion, combining dynamic OCR, advanced QR decoding, and smart web enrichment for seamless integration and analysis.
What's novel
🧠Unlock data insights with VisionFusion, combining dynamic OCR, advanced QR decoding, and smart web enrichment for seamless integration and analysis.
Code Analysis
11 files read · 3 roundsA Streamlit-based pipeline for Iranian businesses that extracts contact data from business card images/PDFs using Gemini OCR and multi-strategy QR detection, web-scrapes extracted URLs for enrichment, translates EN→FA, and outputs merged Excel reports synced to Google Sheets.
Strengths
The QR detection module (qr_dyn.py) implements a genuinely thorough multi-strategy approach with 12+ image preprocessing and detection methods (CLAHE, adaptive thresholding, multi-scale, rotation, pyzbar, zxing fallbacks). The overall pipeline concept—combining OCR, QR, web scraping, and AI translation into a unified business card data extraction tool for the Iranian market—addresses a real niche need.
Weaknesses
Critical security issues: 4 hardcoded Google API keys in source files and a committed secrets.toml. Massive code duplication (scrap.py and excel_mode.py are ~90% identical), a 56kb monolithic app.py, no tests, no package structure, and a broken install script referencing a non-existent app2.py. The README is marketing fluff with nonsensical installation instructions that don't match the actual Python/Streamlit setup.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
729
Contributors
2
Files
21
Active weeks
5
Repository
Language
Python
Stars
1
Forks
0
License
NOASSERTION