Mirra was inspired by how easily gradual changes in someone’s appearance—such as new blemishes, increased redness, under-eye fatigue, or facial puffiness—can go unnoticed from day to day. Existing wellness products often require users to wear and charge another device, manually log information, or remember to take consistent photos. We wanted to create a more effortless experience by turning an existing morning and evening habit—looking in the mirror—into an opportunity to track visible wellness trends.
We built Mirra as an AI-powered wellness mirror with a camera, controlled LED lighting, one-way mirror material, and a Raspberry Pi connected to a paired computer. The camera captures the user’s face while computer-vision software uses OpenCV and MediaPipe to identify facial landmarks, isolate specific facial regions, and compare each scan with the user’s personal baseline. The prototype focuses on detecting possible blemish-like areas, localized redness, and changes in facial fullness. The mirror itself does not contain a screen; results and historical trends are shown separately through the Mirra companion app.
Through this project, we learned that consistent data collection is just as important as the AI model itself. Changes in lighting, camera distance, head angle, and facial expression can significantly affect the results, so we added standardized lighting, positioning guidance, and image-quality checks. We also learned the importance of narrowing the product’s claims. Mirra is designed to track visible wellness changes, not diagnose medical conditions.
Our biggest challenges were capturing a clear image through the mirror material, controlling reflections and lighting, and determining which indicators could realistically be measured with a camera. We initially considered technologies such as LiDAR, but found that they would not provide the color and texture information needed for skin analysis. This pushed us to simplify the prototype and focus on a credible core experience: passive, consistent facial imaging, personal-baseline tracking, and privacy-conscious delivery of useful wellness insights.
Log in or sign up for Devpost to join the conversation.