RecycleVision is an AI-powered Streamlit app that helps users identify types of waste (e.g., plastic, metal, cardboard) and get eco-friendly disposal instructions. It also features a floating chatbot assistant for sustainability tips and recycling guidance — all in a sleek, mobile-ready UI.
✅ Smart Waste Classification
→ Detect waste type from single image, webcam, or batch upload using a MobileNetV2 model.
✅ Eco-Assistant Chatbot
→ Floating WhatsApp-style chatbot powered by OpenRouter's LLM, offering recycling tips and weather-aware disposal advice.
✅ Real-Time Webcam Mode
→ Supports capturing waste via camera for immediate classification.
✅ Batch Mode Support
→ Classify multiple images at once with full label/emoji overlays and tracking.
✅ Dashboard Tracking
→ View number of predictions, class frequency, and eco-stats on the home screen.
✅ Custom Styling
→ A vibrant, mysterious-themed UI with animations, read receipts, and responsive layout.
| Layer | Technology |
|---|---|
| Frontend | Streamlit |
| Backend | Python, NumPy, Pillow |
| ML Model | MobileNetV2 (TensorFlow / Keras) |
| LLM API | OpenRouter(mistralai/mistral-nemo:free) |
| Deployment | Streamlit Cloud |
| Chat Styling | Custom CSS + JS injection |
RecycleVision/ ├── app/ │ ├── chat_bot.py # LLM logic via OpenRouter │ ├── batch.py # Batch image classification │ ├── info.py # Waste info data │ ├── utils.py # Utilities for chat │ └── waste.py # Model loading & prediction ├── model/ │ └── label_map.json # Label mapping for MobileNetV2 ├── pages/ │ ├── Home.py # Dashboard screen │ └── classification.py # Waste classification UI ├── images/ │ └── bg5.jpg # Background for Home ├── main.py # Root Streamlit app with chatbot ├── requirements.txt # Dependencies └── README.md # You're reading it :