en.Wedoany.com Reported - As manufacturing precision increases, industrial quality inspection faces challenges. For instance, microscopic scratches on MPO fiber optic connectors can affect data center link performance, and fine cracks in automotive die-castings pose safety hazards. Manual inspection is inefficient, and traditional equipment struggles with complex curved surfaces and deep cavities.
Guangzhou Inter Intelligent Technology Co., Ltd., incubated from the Guangzhou Research Institute of Xidian University, has partnered with Intel's Industrial Edge Ecosystem to develop a high-precision, high-speed AI visual inspection solution aimed at enhancing defect detection capabilities.
In the optical communication field, specifically for MPO fiber inspection, the company has launched an integrated MPO end-face inspection device. This equipment can simultaneously capture images of four or more end-faces, with a single MPO component inspection taking 4-5 seconds and a per-fiber inspection speed of 0.38 seconds, which is 4-5 times faster than the industry average. Equipped with an auto-focus microscopic module, it can identify scratches and stains as small as 0.5μm, with a missed detection rate below 0.5% and a false positive rate below 2%. One device can replace 4-5 manual workers. AI data accumulation supports quality traceability, linking with front-end process parameters to shift from intercepting defects to preventing them.
For automotive die-casting inspection, the equipment employs a multi-view collaborative imaging system, integrated with an Intel edge server to provide real-time computing power. It achieves 360-degree visual coverage, inspecting areas such as deep cavities and grooves, with a single-point inspection time of 120 milliseconds—74.4% faster than the industry standard. The missed detection rate is as low as 5%, and defect identification accuracy reaches the 0.1-millimeter level.
The company has developed its own universal AI algorithm training platform, utilizing a small-sample, unsupervised training model. Models can be trained with as few as 5 images, and adapting to new components can be completed in 15 minutes, lowering the barrier for production line changes.
As a partner in the Intel Xcelerator ecosystem, Guangzhou Inter Intelligent Technology has already served multiple industry enterprises and plans to expand cooperation in fields such as chips and display panels, promoting AI quality inspection as a manufacturing standard. Through the Intel ecosystem, the company is migrating its technology to more high-end manufacturing scenarios.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
