Dahua Technology Brought Forward Xinghan Large-Scale AI Models

Dahua Technology Brought Forward Xinghan Large-Scale AI Models

Image source: Public Domain

Dahua Technology, a leading global provider of video-centric AI solutions and services, officially launched its Xinghan Large-Scale AI Models, a next-generation, industrial-grade AI system that integrated large-scale visual intelligence with multi-modal and linguistic capabilities. Designed to tackle the complex challenges of real-world environments, Xinghan marked a significant leap in Dahua's ongoing innovation efforts, driving intelligent transformation across multiple industries.

Xinghan Technological Foundation

With the mission of enabling machines to truly understand the world, the Xinghan model system continues to evolve, connecting cutting-edge research with real-world applications. Named after the Chinese word for "galaxy," Xinghan offers a comprehensive capability matrix driven by the synergy between the edge and the cloud, enabling scalable and adaptive intelligence across all industries. Xinghan's enhanced architecture consists of three core model series: L, V, and M. The L series model focuses on natural language understanding and interaction, while the other two address more specific applications:

Series V: Xinghan Vision Models

Centered on advanced visual intelligence and video analytics, this series optimizes target categories by focusing on key targets (e.g., humans, motorized and non-motorized vehicles) to reduce model complexity while maintaining high accuracy.

Main features:

  • Perimeter Protection: Coverage is expanded by accurately identifying smaller targets (even down to 20x20 pixels) compared to traditional CNN-based AI models, reducing false alarms and increasing the detection range of large model cameras.*
  • WizTracking: Offers a state-of-the-art intelligent tracking algorithm that can handle complex occlusions and variations in target pose, achieving a 50% improvement in accuracy.*
  • Crowd Map : Significantly improves detection of small targets at long distances (up to 2x) and includes umbrella compensation, improving accuracy by 80% in rainy conditions*. It also offers a 2.5x increase in scanning range, supports detection of up to 5,000 people, and delivers robust performance in dense crowds and low-light environments.*
  • Scene Adaptive - AI WDR: Leverages situational awareness to analyze the spatial and contextual characteristics of a scene, enabling intelligent, automated camera setup.
  • AI Rules Wizard: Designed for automatic rule delineation for perimeter protection, it offers one-click access, high-precision scene recognition, automatic analysis, and much more.

M Series: Xinghan Multimodal Models

Multimodal models are advanced AI systems capable of simultaneously processing and thoroughly integrating multiple types of heterogeneous data (e.g., text, images, audio, and video). This significantly improves information processing efficiency, enables more natural human-computer interaction, and opens up a wider range of application scenarios.

Main features:

  • WizSeek: Revolutionizes video investigation with natural language search. Simply describe your target (e.g., people, vehicles, animals, or objects) and WizSeek instantly retrieves matching footage from your video archives.
  • Text-defined alarms: Allows users to define alarms simply by describing them in natural language, significantly reducing the development threshold and enabling rapid, flexible, and scalable configuration tailored to various real-world scenarios.