AI for Social Good: Expiry Tracking & Food Donation System

Saving Food. Serving Humanity.

Comment

🥫 AI for Social Good: Expiry Tracking & Food Donation System

🌟 Inspiration

More than 1.3 billion tons of food is wasted globally every year, while millions struggle with hunger and malnutrition.
Local grocery stores and supermarkets discard large amounts of food simply due to expired stock that is not tracked properly.
We realized that a simple AI-powered system could turn wasted food into hope, connecting retail shops with NGOs and people in need.

That thought became our inspiration — to build a solution that reduces food waste and enables social impact using AI and OCR.

💡 What the Project Does

The system helps store owners:

  • Scan product expiry dates using OCR on labels & packaging
  • Automatically track expiry timelines
  • Receive alerts before products expire
  • Sell items at discounted prices or donate to NGOs

The platform uses:

  • PaddleOCR for extracting expiry information
  • ERNIE for generating insights, reports, and intelligent notifications
  • A cloud database for storing scanned inventory
  • A clean web/mobile interface for interaction

🧠 What We Learned

Through this project, we learned:

  • How to process real-world text using PaddleOCR-VL
  • How to convert scanned data to structured formats (Markdown / JSON)
  • How to use ERNIE models to transform and generate content
  • Basics of deploying static content via GitHub Pages
  • Practical experience integrating AI with social-purpose problem solving

We also used mathematical modeling to determine expiry thresholds.
For example, if a product expires in (D) days, notification is triggered when: [ T = D - x ] where ( x ) is the safety margin (e.g., 5 days).

🛠 How We Built It

  1. Created a formatted PDF describing the project problem and solution
  2. Used PaddleOCR doc parser to extract layout & text → Markdown
  3. Sent Markdown to ERNIE-4.5 to generate an HTML webpage
  4. Styled and deployed the result on GitHub Pages
  5. Documented and prepared submission assets

⚠ Challenges We Faced

  • Difficulty configuring OCR environment and handling PDF layout extraction
  • Making the generated HTML clean and responsive with ERNIE prompts
  • Markdown conversion inconsistencies for multi-page PDFs
  • Time spent testing different prompt engineering techniques
  • Understanding which features of ERNIE were most effective for the task

🚀 Future Improvements

  • Real-time expiry tracking mobile app
  • Multi-language OCR for regional grocery stores
  • Direct NGO matching & logistics platform
  • Predictive analytics using ML based on product turnover history

💖 Final Thoughts

We believe that technology should solve real problems.
This project began as a warm-up challenge — but it now feels like something that can create meaningful impact.

"Saving food today means feeding tomorrow."

Built With

Share this project:

Updates