Waste wise

• 

  "Let AI help you on how to dispose of waste"

Inspiration

Most people want to recycle correctly but don't know what goes where. We built WasteWise to close that gap making responsible waste disposal as simple as taking a photo.

What it does

WasteWise classifies waste from a photo into one of six categories cardboard, glass, metal, paper, plastic, or trash and provides practical, step-by-step disposal guidance for each result.

How we built it

The backend is a FastAPI service running a fine-tuned MobileNetV2 model in PyTorch. The frontend is a React + Vite SPA with drag-and-drop upload and live camera capture via the browser MediaDevices API.

Challenges we ran into

A Python environment conflict cost us significant time early on our system uvicorn was spawning workers on Python 3.14 while all dependencies lived in a Conda environment on Python 3.10, causing persistent ModuleNotFoundError crashes that weren't immediately obvious from the traceback.

Accomplishments that we're proud of

We're most proud of the depth of our disposal guidance five researched, actionable steps per category covering contamination risks, sorting requirements, composting alternatives, and specialist collection schemes, rather than a generic one-line tip.

What we learned

Benchmark accuracy doesn't automatically translate to real-world performance. Data diversity varied lighting, backgrounds, and object orientations matters more than model architecture choice.

What's next for WasteWise

Expanding the dataset with TACO, the Kaggle Garbage Classification dataset, and community-contributed images, plus adding new categories like e-waste and organic waste to cover the full range of real-world household items.

Built With

• fastapi
• mobilenetv2
• python
• pytorch
• react
• vite