ChromaShift

Our project was inspired by something personal. One of our teammates is red-green colorblind, and through working with him, we saw how often everyday life depends on color-coded information that many people simply take for granted. Traffic lights, road signs, warning signals, and even small tasks like telling whether a banana is raw or ripe all become harder when color is the main way information is communicated. At night, this problem becomes even more serious, since red against dark backgrounds can be especially difficult to distinguish quickly. That made us ask: how can we use AI and AR to make the world easier to interpret in real time for colorblind users?

We decided to build an accessibility app that acts like a real-time visual assistant. Our solution combines live color remapping with road sign detection. The app shifts colors that are commonly difficult to distinguish — such as reds, blues, and yellows — into more visible alternatives, while also using computer vision to detect road signs and highlight important visual cues. In the future, we envision integrating this directly into Meta Ray-Ban smart glasses, allowing users to receive AR overlays in their field of view without needing to constantly look at a phone.

One of the most important things we did was validate the problem before building. On Day 1, we focused on ideation and customer discovery. We individually interviewed multiple people and asked about how useful this kind of tool would be, whether they would actually use it, and which features would matter most. These conversations helped us realise that the strongest use cases were night driving, road safety, and everyday independence while shopping or navigating public spaces. We also learned that users did not want a complicated interface — they wanted something fast, intuitive, and reliable.

On Day 2, we moved into building the MVP. We implemented the color-shifting functionality and began training a neural network to detect road signs in real time. Our goal was not just to create a visual filter, but to build a smarter assistive system that could identify what actually matters in a scene. On Day 3, we focused on refining the prototype, integrating the object detection into the app, improving the user experience, and preparing our final presentation.

The biggest challenge we faced was balancing ambition with practicality. Real-time computer vision is difficult, especially in low-light conditions, where glare, motion blur, and distance can reduce accuracy. We also had to think carefully about usability: if we added too many overlays or altered the image too aggressively, the app could become distracting instead of helpful. Another challenge was designing for a problem that is highly personal, since colorblindness varies in type and severity from user to user. That forced us to think beyond a one-size-fits-all solution and toward something that could eventually be personalized.

This project taught us that accessibility is not just about adding features — it is about understanding how people experience the world differently, and then designing technology that meets them where they are. In just a few days, we went from a teammate’s lived experience, to customer interviews, to a working prototype that combines AI, accessibility, and AR. More than anything, we learned that the best solutions come from listening first, building second, and making sure the technology serves a real human need.

Built With

• c++
• dart
• flutter
• kotlin
• python
• typescript