IoT-enabled deep learning add-on for dry waste classification with incentivization
Sortify is a smart, retrofittable extension that converts regular dustbins into AI-assisted sorting units. It combines computer vision, embedded hardware, and a mobile experience to classify dry waste as recyclable or non-recyclable, route it into the correct compartment, and reward responsible disposal behavior.
Sortify is designed for campuses and public institutions where dry waste is often mixed due to lack of real-time segregation systems. Instead of replacing existing bins, Sortify adds a detachable hardware module on top of them, making smart waste sorting more affordable and scalable.
- Build a detachable IoT add-on compatible with existing dustbins.
- Classify dry waste into recyclable and non-recyclable categories using a CNN model.
- Add user accountability through QR-based logging and user identity tracking.
- Drive participation through gamification, scoreboards, and monthly rewards.
Sortify is built on the idea that sustainability works best when technology and behavior change are designed together.
- Infrastructure-first practicality: reuse existing bins to reduce deployment cost and friction.
- AI for public good: automate repetitive segregation effort and improve recycling efficiency.
- Behavioral design: make eco-friendly actions visible, trackable, and rewarding.
- Data-driven sustainability: capture usage data for better waste planning and operational decisions.
Sortify targets both environmental and social impact in institutional settings:
- Improves source-level segregation of dry waste.
- Reduces contamination of recyclable streams.
- Encourages consistent user participation through incentives.
- Supports Swachh Bharat goals with measurable campus-level outcomes.
- Creates a replicable framework for smart waste systems in colleges and similar communities.
- User places waste item on/in the Sortify interface.
- Camera captures image.
- Backend model classifies item (
recyclable/non-recyclable). - Arduino + servo mechanism opens corresponding flap.
- Result is shown to the user.
- QR-based action logging updates contribution records.
- User score and leaderboard are updated for incentives.
Sortify uses a two-part mechanical design:
- Primary body: houses electronics (camera module, controller wiring, mounts).
- Flap mechanism: servo-driven sorting interface directing waste by class.
The design was created with parametric 3D modeling to support multiple dustbin sizes and improve manufacturability, assembly, and maintenance.
Direct link to image 1
Direct link to image 2
Direct link to image 3
Based on current project documentation:
- Recyclable classification accuracy: 88%
- Non-recyclable classification accuracy: 84%
- Execution time (proposed setup): ~3.2s to ~3.56s (vs ~6.04s to ~6.56s baseline in report table)
- Python
- TensorFlow + Keras
- OpenCV
- ESP32-CAM
- Arduino Uno
- Servo motors
- OLED/LED display interface
- Camera + supporting circuit components
- Android app (Jetpack Compose, MVVM architecture)
- Firebase Realtime Database
- Firebase Authentication
- Room Database (offline sync)
- QR scanning with ZXing
- Node.js
- JavaScript
- SQL (for structured data handling, per project documentation)
- Data collection and labeling of dry waste images (recyclable vs non-recyclable).
- Image preprocessing (augmentation, resize to fixed dimensions, normalization).
- CNN training with regularization and evaluation (accuracy, precision, recall, confusion matrix).
- Hardware integration with camera capture, model response, and servo actuation.
- User logging, scoring, dashboard updates, and incentive workflow.
Sortify demonstrates that smart waste management does not have to start with expensive replacement infrastructure. A retrofit + AI + engagement model can deliver practical sustainability gains while building long-term environmental responsibility habits in users.