Lumeo Launches Generative Assist to Augment AI Search and Unlock New Training-free Analytics Using Generative AI

Key benefits of Generative Assist include:

• •Flexible Training-free Analytics: Generative Assist uses vision language models to allow users to utilize natural language prompts to ask questions or extract scene and object attributes similar to human observation. This capability enables new analytics use cases and integrations that previously required custom models or were difficult to develop.
• •Smarter Captions and Alerts: By using large vision models, the platform enhances the quality of alerts and scene descriptions. Users will receive detailed, context-rich insights, transforming generic notifications into actionable intelligence. For example, alerts can specify, "Two people detected wearing masks and attempting to open the door," improving incident response efficacy.
• •Augmented AI Search: Generative Assist indexes objects, attributes, and custom event descriptions, significantly improving the speed and precision of video retrieval. The description and attributes augmented by Generative Assist feed into Lumeo's AI Search capability, thereby allowing users to conduct searches through their video footage using natural language queries, such as: "People wearing ski masks." "Locate any bicycles in the videos." "Were there any fights in the footage?" "Is the dumpster overflowing?"

Generative Assist works by integrating vision language models such as OpenAI GPT-4, Google Gemini, Nvidia NIMs, Llama through APIs. To manage the intensive computational requirements of large vision models, Generative Assist uses conditional or periodic processing — enabling users to lookup an object or scene based on various conditions such as a new person entering a scene or a vehicle stopped for a long duration. Additionally, Generative Assist can run on the edge (on high GPU capacity servers) or in the cloud for devices with constrained memory or compute. This hybrid approach optimizes bandwidth and resources, resulting in cost efficiencies.

By integrating these cutting-edge features, Generative Assist equips organizations to derive actionable insights from their video data more efficiently, facilitating informed decision-making in sectors such as security monitoring, retail analytics, and industrial optimization.