Edge application management method and system based on 5G + AI industrial vision

Abstract

The invention relates to the technical field of edge computing, in particular to an edge application management method and system based on 5G + AI industrial vision. The efficiency and quality can be improved for enterprises, and the cost is reduced; The method comprises a cloud platform ME orchestrator MEO, the cloud platform ME orchestrator MEO receives an uploaded ME instance packet, the ME instance packet further comprises a deployment mode indication, and the deployment mode indication adopts one of multiple deployment modes.

Description

technical field [0001] The present invention relates to the technical field of edge computing, in particular to an edge application management method and system based on 5G+AI industrial vision. Background technique [0002] Under the guidance of "new infrastructure", technologies such as 5G and AI are being implemented. 5G's unique features such as large bandwidth, low latency, and high reliability, combined with the "intelligent" features of AI, are becoming the industry's digital transformation driving force. [0003] As the application of industrial machine vision becomes more and more extensive and the objects to be detected become more and more complex, the application of machine vision is transitioning from traditional industrial vision to AI industrial vision based on deep learning. [0004] Although AI industrial vision based on deep learning can solve more complex problems and is more and more widely used, there are still the following problems that restrict faste...

AI Technical Summary

Problems solved by technology

[0004] Although AI industrial vision based on deep learning can solve more complex problems and is more and more widely used, there are still the following problems that restrict faster development:

[0005] The cost of end-side computing power is getting higher and higher; as the pixels of industrial cameras become higher and higher, and the application of deep learning in industrial visual inspection, the requirements for computing power hardware performance are getting higher and higher, and the application of high-performance CPU / GPU is also increasing. It is becoming more and more widespread, which directly greatly increases the cost of computing power. At the same time, a set of vision systems often deploys a computing power device independently. All computing power cannot be shared and reused, and the utilization rate often cannot reach the highest level.

[0006] The maintenance cost of a single point system is too high; the deployment method of a single machine with a fixed position requires that each device debugging and software update maintenance need to be completed locally. This kind of maintenance cost will be too high in large-scale application scenarios

[0007] Data islands, data cannot be effectively shared; in some industrial vision systems, image processing is done locally and cannot be shared between systems in a timely manner, which will affect production progress

The construction cost of wired Internet in the factory floor is too high and the capacity is limited

Images