SightLine

We’re not building just another camera app. We’re building a visual intelligence system. Our goal is simple: Intelligence in Every Glance.

Comment

Inspiration

The inspiration for SightLine came from the realization that "seeing" isn't the same as "understanding." For the 2.2 billion people globally with vision impairment, or even for travelers in a foreign country, the world is filled with visual noise that is difficult to decipher. We wanted to build a tool that doesn't just describe a scene, but actually deciphers it, prioritizing what matters most to a user's safety and immediate goals.

What it does

SightLine is a tactical visual intelligence system. It uses a real-time camera feed or image uploads to "read" the environment. Using advanced AI, it identifies key entities like transit schedules, menus, and safety hazards. It then categorizes them, prioritizes urgent alerts, and can translate information into around 6 languages. It also features a "Neural Watchlist" that allows users to set custom triggers for specific objects or text they are looking for.

How we built it

We built the application using a modern, high-performance stack: Frontend: React 19 and TypeScript for a robust, type-safe user interface. Intelligence: Integrated Gemini 1.5 Flash via the @google/genai SDK to perform multimodal analysis, allowing the app to understand both visual and textual context simultaneously. Styling: Tailwind CSS 4.0 was used to create a unique "Tactical HUD" aesthetic. Animations: Framer Motion (motion/react) powers the smooth, high-tech transitions and the "scanning" visual effects.

Challenges we ran into

One of the biggest challenges we faced was getting the upload feature to function properly. Our project allows users to upload images, and while we were able to successfully upload them, the images were not being analyzed afterward. It took us nearly three hours to resolve this issue, even though it seemed small. However, we couldn’t give up because this feature was a major part of our project. We eventually realized it was a state-synchronization issue between the file reader and the AI processing hook, and overcoming this was a huge turning point for the team.

Accomplishments that we're proud of

The Tactical UI: We are incredibly proud of the "Mission Control" aesthetic we achieved, which makes the app feel like a professional-grade tool rather than just a demo. Neural Extraction: Successfully prompting the AI to return structured, prioritized JSON data from raw images with extremely low latency. Watchlist Logic: Implementing a system that can cross-reference real-time visual data against a user-defined list of interests.

What we learned

We learned a tremendous amount about multimodal AI and how to handle large data streams in React. Specifically, we gained deep experience in "Prompt Engineering" for vision models, learning how to guide the AI to ignore visual "noise" and focus only on actionable data. We also learned the importance of persistence when debugging complex asynchronous features like our image upload system.

What's next for SightLine

The next step for SightLine is moving beyond the smartphone screen. We envision integrating this technology into AR Smart Glasses to provide a persistent, hands-free heads-up display. We also plan to implement a Neural Audio Link, using spatial audio to "whisper" the location of identified objects to the user, creating a truly immersive accessibility experience. The final thing we would want to do is to add more languages so we can be more diverse and be used across the world.

Built With

Share this project:

Updates