In today's hyper-visual world, enterprises are amassing video content at an unprecedented rate. Yet, this explosion of visual data has created a significant challenge: the majority of it remains "dark data"—unstructured, unsearchable, and largely untapped. Finding a specific event, object, or spoken word within terabytes of footage is a monumenta