♻️ Inspiration

"Wishcycling" is a major issue: people often throw items into the recycling bin with good intentions, but incorrect sorting contaminates entire batches. Modern packaging is multi-material (for example, a bottle, cap, and label made of different plastics or metals), and recycling rules vary significantly by region. We built SortSense to solve this confusion and give citizens absolute clarity on how to sort every single component.

⚙️ What it does

SortSense allows users to:

Scan barcodes: Instantly fetch packaging components (bottle, cap, label, etc.) using Open Food Facts or offline demo data[cite: 7].
Search everyday items: Use a fuzzy search knowledge base for items without barcodes (like pizza boxes, batteries, or coffee cups)[cite: 5].
Switch regional rules: Adapt verdicts instantly between regional guidelines (e.g., France vs. United States single-stream rules)[cite: 8].
Challenge Mode: Test sorting knowledge and log mistakes before seeing the correct bin[cite: 12].
Impact Tracking: Track stats on items sorted, mistakes avoided, and estimated CO2e diverted via a lightweight SQLite backend[cite: 9].

🛠️ How we built it

Backend & API: Python with Flask, featuring robust normalisation of messy packaging tags and rules (rules.py, off.py, kb.py)[cite: 6, 7, 8].
Database & Metrics: SQLite (impact.py) to log sorting events and calculate diverted mass and avoided mistakes[cite: 9].
Frontend: Responsive single-page UI (index.html) built with modern CSS and vanilla JavaScript, including barcode scanner support via the Web API[cite: 12].
Testing: Comprehensive test suite using pytest and hypothesis for core logic and edge cases[cite: 11].

🚧 Challenges we faced

Standardising messy and inconsistent packaging metadata across thousands of products into clean, actionable material categories (plastic, glass, metal, paper, carton).
Designing a flexible rule engine capable of handling regional exceptions (e.g., plastic film rules differing between France and the US)[cite: 8].

🚀 What's next

Integrating photo-based AI recognition for unbarcoded items or damaged packaging.
Connecting official municipal rule feeds.
Establishing store-side partnerships for specialized drop-off locations (e.g., batteries, electronics, and soft plastics).