Cabinet intelligent asset inventory method based on code recognition

Abstract

The invention discloses a cabinet intelligent asset inventory method based on code recognition, and the method comprises the following steps: 1, preprocessing a code label image; 2, constructing and improving a PSENet network model, wherein the model comprises a feature extraction module, a segmentation head module and a post-processing module; 3, recognizing asset tags with different binary numbers, obtaining asset information, wherein a picture recognition model needs to be trained through a convolutional neural network before identification, and the effect of returning numbers through a given graph or pattern is achieved; 4, obtaining the database stores asset information; and step 5, summarizing asset data, and returning inventory information. According to the invention, multiple pieces of identification information can be captured at a time, a unique binary code exists after identification, and the management requirement for assets corresponding to labels in a natural scene is met.

Description

technical field [0001] The invention relates to the field of asset inventory, in particular to a method for intelligent asset inventory by PSENet network-based migration learning code recognition for hardware neatly placed in a server cabinet. Background technique [0002] Today, due to the continuous improvement of the demand for innovative technology development in the Internet industry, my country's data center application scale has reached the second place in the world. The continuous increase in the utilization rate of data center equipment and network cabinets also means that the number of fixed assets continues to increase. Improving the efficiency of fixed asset management is an effective way to ensure the utilization rate of fixed assets. [0003] Traditional data center equipment management methods that use manual entry or tools such as Access, Excel, or small asset management software often have high labor costs, high error rates, low efficiency, and often unavail...

AI Technical Summary

Problems solved by technology

[0003] Traditional data center equipment management methods that use manual entry or tools such as Access, Excel, or small asset management software often have high labor costs, high error rates, low efficiency, and often unavailable data

The relatively mature asset barcode scanning technology can quickly enter information by scanning labels or QR codes on assets, which improves the shortcomings caused by manual methods and improves work efficiency. Long working hours, and the core of the RFID technology that can be automatically identified is the RFID tag, and its instability makes it easy for assets to be missed

There is also a method of inventorying assets through image recognition. Although the information can be entered more accurately, usually the fixed assets of the data center are neatly placed in the cabinets. It will be inconvenient to obtain images when there are a large number of them.

[0004] Today, with the continuous development of artificial intelligence, many excellent algorithms are constantly being born in the field of target detection. Most of the algorithms are for the detection of objects and organisms. The targets are relatively large, and text labels are usually included in a picture. It is presented in a relatively small form, and there is a bending situation, and the information of the original label is not only a requirement to be detected, but the extraction of label information is more important in the actual scene. In this case, the general target detection is obviously is not enough

Images