Takkubi

The Problem: While hardware like Neuralink's Blindsight is in clinical trials, the software pipeline to translate real-world visual data into sensory signals is needed today. ​The Solution: NeuroSight acts as a real-time Brain-Computer Interface (BCI) simulator. It uses a custom OpenCV "Center-Gravity Target Lock" to dynamically track objects, extracting structural density, color, and texture. ​How it works: This spatial data is instantly sent to Meta Llama 3.2, which generates a contextual sensory description. This is then fed to the user via the Web Speech API for low-latency cognitive audio guidance. ​How I built it: I built the backend using Python, FastAPI, and OpenCV for real-time computer vision. The AI translation is powered by Meta Llama 3.2 via the Groq API, and the frontend is a lightweight HTML/JS interface ensuring fast execution without heavy hardware requirements.