TWM637375U - Three-in-one artificial intelligence defect detection model training and detection platform - Google Patents

Abstract

A three-in-one artificial intelligence defect detection model training and inspection platform, comprising a sampling and inspection unit, a server, a web page platform, and a three-in-one visualization information system, wherein the server has a plurality of artificial intelligence defect detection models, a web platform, and a three-machine integrated visualization information system. The three machines are a computer, a mobile phone, and a tablet. The web platform can operate the artificial intelligence defect training and detection model, and control a sampling and inspection unit to conduct images of the inspected objects. Samples are sent to the server for model training and testing, and the testing results are presented through the web page in a three-in-one visual information system, so that the AI defect detection model training and testing can be performed on the web platform in the same environment without being affected by the operating system and for the purpose of hardware device restrictions.

Description

Three-in-one artificial intelligence defect detection model training and detection platform

This creation is about a three-in-one artificial intelligence defect detection model training and detection platform, which provides users with the use of the web platform for defect sampling and photography, database archiving, and control of artificial intelligence detection model training and detection, and uses the three machines of the web page The integrated method provides users with viewing, achieving the purpose of remote artificial intelligence defect detection model training and detection.

With the automation of industrial production lines, smart manufacturing and product inspection have become more important. In addition to reducing the outflow of defective products, it can more accurately inspect defects and reduce labor costs. General artificial intelligence defect detection usually needs to first collect defects and classify and file them. After artificial intelligence training, it is applied to the actual production line for defect identification, and then a fixed screen connection is used for users to view the results.

This patent develops a three-in-one artificial intelligence defect detection model training and detection platform, mainly for the purpose of collecting and sampling defects, data archiving, training and detection by connecting the detection unit, database and artificial intelligence defect detection model in series on the web platform. The concept of webpage combined with three machines is used to realize the presentation of visual information system, so that the artificial intelligence defect detection model training and detection can be carried out in the same environment on the webpage platform, and it is not affected by the operating system and hardware installation. for the purpose of restriction.

100: Sampling and detection unit

101: Lens module

102: Sampling and testing device

103: Products to be inspected

200: server

201: Database

202: Temporary memory

300: Web Platform

301: Classification of sampling defects and moving folders

302: Defect labeling function

303: Automatic operation and parameter adjustment of image amplification

304:Required conversion for conversion of artificial intelligence defect detection model

305: Training model

306: View the training model and model file after training

307: Detection system

400: Three-in-one visual information system

401: computer screen

402: Tablet

403: mobile phone

Figure 1 is a schematic diagram of the three-in-one artificial intelligence defect detection model training and detection platform created by this creation.

Refer to Figure 1, which is a schematic diagram of the three-in-one artificial intelligence defect detection model training and detection platform created by the author. The operation method is divided into (1) web platform driven sampling and detection unit sampling product defects to be inspected, (2) web platform driven data The data is imported into the database for archiving, (3) the web platform drives artificial intelligence defect model training and detection, (4) the web platform drives the sampling and detection unit to perform defect detection.

The user places the product 103 to be inspected on the inspection device 102 of the inspection unit 100, and controls and drives the lens module 101 through the web platform to collect and sample defects, feed them back to the database 201 of the server 200, and present them on the web page for archiving. On this webpage platform 300, users can perform sample defect classification and move folders 301, defect labeling function 302, automatic operation of image augmentation and parameter adjustment 303, transfer required for converting artificial intelligence defect detection models 304, Training model 305, post-training model file viewing 306, after complete sampling, conversion, amplification, and training, this platform can be used for defect detection operations, and the detection system 307 is used to control model files, parameter settings, detection thresholds, and a detection unit for detection. The test result screen will be stored in the temporary memory 202 database. After waiting for the program to retrieve the test result image, the continuous screen will be presented in the form of the three-in-one visual information system 400. This interface will be corresponding to the screen size of the device used. Self-adaptive, users can choose to use three machines at will, namely computer screen 401, tablet 402, and mobile phone 403 The device obtains the screen.

Although this creation has been disclosed as above with a preferred embodiment, it is not intended to limit this creation. Anyone who is familiar with this skill can make various changes and modifications without departing from the spirit and scope of this creation. Therefore, this The scope of protection of a creation shall be defined in the scope of the attached patent application.

100: Sampling and detection unit

101: Lens module

102: Sampling and testing device

103: Products to be inspected

200: server

201: Database

202: Temporary memory

300: Web Platform

301: Classification of sampling defects and moving folders

302: Defect labeling function

303: Automatic operation and parameter adjustment of image amplification

304:Required conversion for conversion of artificial intelligence defect detection model

305: Training model

306: View the training model and model file after training

307: Detection system

400: Three-in-one visual information system

401: computer screen

402: Tablet

403: mobile phone

Claims (5)

A three-in-one artificial intelligence defect detection model training and detection platform, including: a sampling and detection unit; a server; a webpage platform; and a three-in-one visual information system; three-in-one artificial intelligence defect detection model training and detection The platform builds the web platform and the three-in-one visual information system on the server. Through the artificial intelligence defect detection model in the server, with the webpage control artificial intelligence as the main axis, the defect detection is carried out on the product to be inspected on the sampling and detection unit. And the detection results are presented in real time with the three-in-one visual information system, so as to achieve the purpose of using the web platform to conduct artificial intelligence defect detection model training and testing, and the three-in-one result presentation and operation without being limited by the operating system and hardware devices. The three-in-one artificial intelligence defect detection model training and detection platform as described in Claim 1, wherein the detection unit includes: a detection unit, a lens module, a sampling and detection device, and a product to be inspected. The three-in-one artificial intelligence defect detection model training and detection platform as described in claim 1, wherein the server includes: a database and a temporary storage memory. The three-in-one artificial intelligence defect detection model training and detection platform as described in claim item 1, wherein the web platform includes: sampling defect classification and mobile folder, defect labeling function, automatic operation of image amplification and adjustment parameters, conversion The required conversion of the artificial intelligence defect detection model, training model and post-training model file viewing, and webpage detection control, including model files, parameter settings, detection thresholds, and detection units including cameras and lenses. The webpage platform is the main Control and serial integration of the overall three-in-one artificial intelligence defect detection model training and The main core of the detection platform is the purpose of integrating software and hardware, convenience, speed, filing, classification, real-time presentation and not being limited to any system. The three-in-one artificial intelligence defect detection model training and testing platform as described in claim 1, wherein the three-in-one visual information system includes: a computer screen, a tablet, and a mobile phone, and the three-in-one device can be connected to the Internet The visual information system in the server provides an adaptive interface according to the corresponding device.

Images