team CodeCycle

Inspiration

The growing issue of recycling contamination—caused largely by misidentified waste—was the inspiration behind sortAIble. Existing solutions often fall short in real-world contexts due to limited accuracy and clunky interfaces. I set out to develop a streamlined AI-driven platform that would make waste sorting effortless, educational, and effective for everyone.

What it does

The app leverages a computer vision model trained to identify plastic resin codes from real-world images. Once an image is uploaded or captured, sortAIble analyzes it and delivers actionable disposal guidance based on the material class—empowering users to make the right environmental choice in seconds.

How we built it

sortAIble is built using Python, TensorFlow, JavaScript, and CSS. The core model is a custom-trained convolutional neural network, refined across diverse plastic waste samples. The front-end is lightweight, responsive, and designed to simplify user interaction without compromising on detail or speed.

Challenges we ran into

A key challenge was the imbalance in dataset distribution—some resin types (like Class 5 plastics) appeared disproportionately, leading to biased predictions. Attempts at class weighting and data trimming initially worsened accuracy. Eventually, I restructured the dataset and fine-tuned the training process to achieve a balanced, high-performing model.

Accomplishments that we're proud of

• Developed a reliable waste classification model with high real-world accuracy
• Enabled resin-code detection even under poor lighting or partial occlusion
• Built a seamless interface bridging machine learning and user experience
• Delivered instant, localized recycling guidance with minimal input friction

What we learned

Through this project, I gained deep insight into dataset balancing, real-world model testing, and the intersection of AI and sustainability. It reaffirmed the value of designing with purpose and showed how AI can be a tool for impactful change when aligned with everyday habits.

What's next for team CodeCycle

The roadmap for sortAIble includes:

• Mixed-material detection for composite waste types
• Full mobile deployment with camera integration and offline access
• Multilingual support to promote inclusive, global accessibility
• Partnerships with recycling organizations to align guidance with local regulations
• Gamified learning features to boost long-term environmental engagemen

Built With

• css
• html
• javascript
• python
• tensorflow