Hikvision unveils Guanlan Large-Scale AI Models to power next-gen AIoT products and applications

HANGZHOU, China, April 29, 2025 /PRNewswire/ -- Hikvision has announced the official launch of its Guanlan Large-Scale AI Models, a suite of machine learning engines pre-trained on deep industry knowledge. Designed to excel in computer vision, natural language processing, and multimodal fusion, these models set a new benchmark for AI-powered Internet of Things (AIoT) applications.

The philosophy behind Guanlan

Inspired by an ancient Chinese philosophy that reads the subtle language of water—"To understand the nature and movement of water, one must observe its waves"—Guanlan embodies the art of perceiving the underlying nature, principles, and nuanced dynamics of all things. Derek Yang, Vice President of Hikvision International Business Center, remarked: "Guanlan's core philosophy of uncovering hidden patterns aligns perfectly with our mission to deliver smarter, faster, and clearer insights through large-scale AI models."

Three-tier architecture driving scenario-based digitalization

Hikvision's Guanlan Large-Scale AI Models feature a three-tier architecture, consisting of foundation models such as large vision, language, and multimodal models, industry models at the intermediate layer, and task models at the top. This architecture is designed to meet a wide range of needs from foundational functionality to complex scenario applications, enabling unprecedented capabilities and usability with:

• Multimodal perception abilities that combine images, text, and more
• Multiple deployment methods for advanced, flexible applications

With companies increasingly adopting AIoT technologies to address specific operational challenges and accelerate digital transformation, Hikvision has been shifting its focus from video security to broader AIoT applications. On this path, it has become increasingly clear that large-scale AI models will play a critical role in revolutionizing the industry, driving Scenario Digitalization across public and private sectors.

Product highlights

• DeepinViewX Cameras: Powered by Guanlan's large vision models, these cameras double video content analysis (VCA) range, halve repeated alarms, boost detection rates, and cut perimeter false alarms by over 90% compared to conventional AI solutions.
• AcuSeek NVRs: Guanlan's large multimodal models allow users to search for specific objects in video footage using natural language queries. Hikvision's AcuSeek NVRs simplify the process of video retrieval, making it faster and more efficient.
• Expanded AIoT Portfolio: Upcoming deployments include HeatPro-Series Thermal Cameras and Intelligent Checkpoint Cameras—both enhanced by Guanlan's unparalleled agility and precision.

Try Guanlan today

Large-scale AI models greatly enhance object detection and information retrieval in videos. We look forward to seeing more industries benefit from Hikvision's Guanlan Large-Scale AI Models and enjoy the intelligence and convenience they provide. We aim to promote the implementation of Guanlan Large-Scale AI Models across a variety of industries and organizations.

To learn more, visit our AIoT Technologies webpage. Alternatively, contact your regional Hikvision representative to schedule a live demo and discover how Hikvision products with Guanlan Large-Scale AI Models can transform operations—on demand and at scale.

source: Hikvision Digital Technology