A local, low-latency system that uses Vision-Language Models (Moondream 2 & Florence-2) to generate natural language descriptions of live camera footage. Built for accessibility, content creation, ...