Neurolens

Inspired by the gap between gut instinct and real emotional response, we built a real-time webcam analytics tool that detects joy, surprise, boredom, and more using AI. Over 24 hours, we combined React and Vite for a sleek frontend with live video and dynamic charts, and a FastAPI backend that processes frames, crops faces (or falls back to full frames), and queries Hugging Face’s emotion model via its Inference API all while prioritizing privacy with in-memory processing and user consent. We battled Windows compatibility issues, finicky face detection in low light, and cryptic Hugging Face API errors (like the infamous 'NoneType' object has no attribute 'lower'), but learned to embrace pragmatic fixes: skipping local face cropping when needed, validating every image byte, and tuning OpenCV aggressively. In the end, we shipped a visceral, demo-ready experience that turns fleeting micro-expressions into actionable insights—proving that even in a hackathon, empathy can be engineered.