AI video analysis and identification system
Keywords: AI video behavior recognition and analysis system, AI visual analysis system, AI image recognition and analysis system, AI recognition system, AI behavior analysis system
Version: 1.20.4.3
一.summary
AIBOX algorithm system in the core technical framework of edge computing terminal, x86 server, domestic letter server, the realization of customer needs as the goal, to enable the customer big data platform AI. The AI algorithm middleware system with high accuracy and strong business adaptation ability can provide powerful technical ns for social security management and safe production.
二.system architecture
2.1 System network topmap
2.1.1 Typical application of LAN mode
This mode is a typical application in the LAN environment. All devices are connected to the same LAN. The AI system server can access to various brands of webcam and hard disk video recorder, and can also connect to some alarm devices and sensors.
Client management computer logs in to the main server and manages the system in the user interface of the client software.
One main AI server supports access to multiple slave AI servers to realize stack access management and facilitate system expansion.
Diagram representation of the LAN application
2.1.2 Centralized Internet management of multi-division local AI system
Note: This mode is multiple branch AI systems scattered around the country, and alarm events are uploaded to the cloud platform and connected to the cloud server through the IE client. The local AI system architecture is LAN mode, as long as the local AI server is set to the cloud management server address information. Features of this mode: two-level frame color, local local unit system (local operation), the central system unified the alarm record of the local system of all divisions
Internet edge end analysis + cloud centralized management application diagram
2.1.3 Internet background analysis mode
Graphram of Internet backend analysis
Note: The video sources distributed in various places will be uploaded to the cloud streaming media server through the national standard GB28181 protocol, and then provide the back-end AI server analysis and identification of the streaming media protocols.
2.2 AI-MIS algorithm middleware
2.2.1 Introduction to the middleware of the AI-MIS algorithm
AI-MIS adopts the underlying architecture of artificial intelligence machine vision "deep learning" technology, combined with application-level algorithm, and adopts component design to facilitate the expansion of algorithm types. At the same time, it has strong sample processing ability and sample automation production label capability, and can efficiently develop new algorithm components.