Clean Vision

Live feed POV
Dashboard with scanned objects
Physical Glasses
Human wearing glasses

Inspiration

We were inspired to create a cost-effective and efficient solution to reduce energy waste in everyday environments such as homes, offices, and other high-traffic spaces. Many people overlook small sources of energy loss because they don't realize the financial impact over time. By providing users with real-time estimates of energy costs and inefficiencies, we hope to encourage smarter energy habits and promote the adoption of more sustainable solutions, such as solar power.

What it does

Clean Vision uses real-time data captured by the Meta Aria Research Glasses to analyze a room and perform an automated energy audit. Using AI-powered object detection, it identifies appliances and devices, estimates their energy consumption, and detects potential sources of energy waste. The platform also incorporates AI-based authentication to ensure secure access and personalized reports.

How we built it

We built a real-time dashboard that processes data captured by the Meta Aria Research Glasses to identify objects within a room and estimate their power usage. The collected information is then sent to Google Gemini, which generates a detailed energy audit highlighting estimated energy costs, potential energy leaks, and personalized recommendations for improving efficiency and reducing electricity bills.

Challenges we ran into

One of our biggest challenges was integrating Google Gemini with our real-time computer vision pipeline while maintaining fast performance. Bridging hardware, software, AI models, and live sensor data into one seamless workflow required significant debugging and optimization. Ensuring accurate object detection while minimizing latency was another major obstacle throughout development.

Accomplishments that we're proud of

We are especially proud of how much our object detection system improved during development. Early versions frequently misidentified objects, such as confusing monitors with televisions or missing devices entirely. Through continuous testing, refinement, and tuning, we greatly increased the accuracy of our detection pipeline, resulting in much more reliable energy audits and recommendations.

What we learned

Throughout this project, we learned how important rapid iteration and continuous refinement are when developing AI-powered products. We also gained valuable experience integrating external large language models into a real-time application while overcoming compatibility and debugging challenges. In addition, we learned best practices for securely managing API keys and protecting sensitive credentials to prevent unauthorized access or data leaks.

What's next for Clean Vision

Our vision is to make Clean Vision accessible to everyone. We plan to develop a mobile application that allows users to perform an initial AI-powered energy audit using only their smartphone camera. For more comprehensive inspections, energy professionals could use the Meta Aria Research Glasses along with thermal cameras and additional environmental sensors to generate even more accurate measurements of energy consumption and energy loss. By providing precise estimates of potential savings, Clean Vision can help homeowners make informed decisions about energy-efficient upgrades, including solar panel installations, while giving them a clear understanding of how much money and energy they can save over time.