Architecture of computer vision processing server based on neural network

Abstract

The invention discloses a computer vision processing server structure based on a neural network, comprising a parent server and a plurality of child servers. Each child server is attached to the parent server at the same time. Each sub-server is provided with a neural network, and each sub-server performs preprocessing through its own neural network at the same time. The parent server further processes the preprocessed image data to form the computer vision processing data. The invention discloses a computer vision processing server structure based on a neural network. Firstly, a sub-server preprocesses pictures or videos requiring computer vision processing, thereby extracting image features, realizing data desensitization and avoiding privacy leakage problem. Then the parent server processes the pre-processed image data in depth, so that the server architecture can achieve load balance as far as possible, reach the balance point, and realize the maximum utilization of each server 'sresources.

Description

technical field [0001] The invention belongs to the technical field of servers, and in particular relates to a neural network-based computer vision processing server architecture. Background technique [0002] Concentrate the application system website, database, and file system on the same server, thus forming the most elementary server system. However, the above-mentioned server system bears too much pressure, and the bandwidth, CPU, and memory pressure may also be too large, and there is a risk of downtime. [0003] On the other hand, current AI projects are more computationally intensive. These projects usually use GPU servers, which also have similar problems, making it difficult to achieve load balancing among multiple servers. [0004] In addition, if image or video processing based on computer vision needs to be uploaded to a server for calculation and processing, the transmission process may also involve the issue of image privacy disclosure. Contents of the inv...

AI Technical Summary

Problems solved by technology

However, the above-mentioned server system bears too much pressure, and the pressure on bandwidth, CPU, and memory may also be too high, and there is a risk of downtime

[0003] On the other hand, current AI projects are more computationally intensive

These projects usually use GPU servers, which also have similar problems, and it is difficult to achieve the purpose of load balancing between multiple servers

[0004] In addition, image or video processing based on computer vision, if it needs to be uploaded to the server for calculation and processing, may also involve the issue of image privacy disclosure during the transmission process