sortAIble is an intelligent waste classification tool that uses deep learning to recognize plastic resin codes from images—instantly guiding users on how to properly dispose of plastic waste. Built to reduce recycling contamination and make sustainability simple, sortAIble brings the power of AI to everyday environmental action.

Recycling contamination remains a serious global issue. Improperly sorted plastic often renders entire recycling batches unusable. Existing tools are either too technical or inaccurate in real-world conditions. This inspired me to build sortAIble—a lightweight, user-friendly app that uses AI to empower users with immediate, clear disposal guidance, one scan at a time.

• Classifies plastic waste using computer vision trained on real-world resin code images
• Detects resin types like PET, HDPE, and PP with over 90% accuracy
• Provides real-time, location-neutral disposal guidance
• Offers a simple, intuitive interface for quick interaction

• Python – core logic and model training
• TensorFlow – deep learning and image classification
• JavaScript – dynamic front-end interaction
• CSS & HTML – responsive, clean UI/UX design

I trained a convolutional neural network (CNN) on a curated dataset of plastic waste images with annotated resin codes. The model was refined to perform well despite noise, partial codes, and lighting variability.

• Addressed data imbalance (e.g. Class 5 plastics were overrepresented)
• Iteratively improved performance with class weighting and smart dataset pruning
• Achieved balanced model predictions across all target resin types

• Real-time image-based classification of plastic waste
• Visual detection even with damaged or faint resin symbols
• Streamlined user interface for rapid interaction
• Encouraged sustainable habits through accessible AI

• Advanced model tuning under real-world constraints
• Practical application of AI in sustainability contexts
• Designing usable AI tools with inclusive and scalable UX principles

• 📱 Mobile app integration with camera and offline support
• 🌍 Multilingual user interface
• ♻️ Mixed-material and non-plastic waste classification
• 🤝 Collaboration with local recycling authorities
• 🏆 Gamified learning for better recycling behavior

Got suggestions, ideas, or want to contribute?
Open an issue, drop a ⭐️, or reach out—I’d love to hear from you.