Vision Language Model (VLM) Video Search from Ipsotek

6th January 2025, London – Ipsotek, an Eviden business and global leader in AI Computer Vision solutions, has today announced the launch of VLM, a groundbreaking addition to its VISuite platform that redefines real-time video analytics with Vision Language Models (VLMs). VLM represents a major advancement in Generative AI integration, using VLMs to achieve detailed scene understanding and contextual insights empowering operators to make informed decisions promptly.

Building on Ipsotek’s Natural Language Detector introduced in April 2024, VLM allows users to ask questions like, “Let me know when something unusual is happening in any camera view” and receive a detailed response describing the unusual aspect of the captured behaviour. For instance, it might respond, “Yes, there is a flood; water levels are rising in the northern section, and several vehicles are stranded, causing heavy traffic congestion,” providing actionable insights that enable quick decisions.

Key Advantages of VLM:

• Zero Configuration Deployment: VLM eliminates predefined alarm configurations, intuitively understanding scenes and ensuring seamless scalability across thousands of cameras.
• Contextual Accuracy: By answering generic questions, VLM delivers highly accurate insights, providing operators with critical details beyond the traditional alarms with static descriptions.
• Scalability: VLM seamlessly scales to support vast deployments, making it suitable for industries ranging from transportation to retail and beyond.

“VLM is a game-changer for video analytics. By combining Computer Vision with the power of Vision Language Models, we’ve empowered users with an unprecedented level of interaction with Ipsotek’s VISuite platform that led to intuitive experience with enhanced benefits,” said Ipsotek CTO & Sales Director Dr. Boghos Boghossian. We are very excited about the future advancements this innovation will bring.’’

Ipsotek’s VISuite platform is already tailored to over 17 industry verticals, and future updates will bring even more specialised capabilities.

A Vision for the Future

VLM is just one milestone in Ipsotek’s ambitious roadmap to bring Generative AI to the forefront of AI Video Analytics. Future developments, set for release in 2025, will further transform how operators interact with and gain more value from Computer Vision systems. Ipsotek’s R&D team is actively working on future VLM features aiming to design tools that elevate real-time decision-making, operational safety, and automation to the next level.