LG Innotek, a global manufacturer of materials and components based in South Korea, is working with Intel to implement an AI-powered smart factory at its Gumi production facility. The initiative uses Intel Core and Xeon processors, as well as Arc discrete graphics processing units (GPUs), integrated through the OpenVINO software toolkit.
LG Innotek has traditionally used rule-based and deep learning inspection systems on its production lines to ensure product quality. Now, the company aims to automate inspection throughout every stage of manufacturing using artificial intelligence.
Intel’s solution involves streaming data from the production line into PCs powered by Intel Core CPUs for defect analysis. More demanding tasks, such as running multiple algorithms on high-resolution images, are handled by Intel Arc discrete GPUs. Accumulated datasets are then sent to servers equipped with Intel Xeon processors for pre-training purposes.
The collaboration may expand in the future to include servers with Intel Gaudi AI accelerators for managing pre-training workloads. This combination of technologies has reduced the cost of building AI inspection systems for LG Innotek.
According to the companies, switching to Arc-based discrete GPUs in LG Innotek’s main manufacturing site has improved cost efficiency compared to other hardware solutions. These savings have allowed LG Innotek to consider scaling up its use of AI vision inspection across more production lines domestically and internationally.
Last year, LG Innotek applied Intel’s AI vision inspection technology to its mobile camera module production line. This year, it plans further expansion at major domestic sites like the Gumi 4 factory—which produces flip-chip ball grid arrays—and overseas facilities.
Engineers at LG Innotek initially hesitated about integrating built-in GPUs due to concerns over compatibility with their existing deep-learning environment. However, they found that OpenVINO made this transition straightforward: “Since OpenVINO launched in 2018, Intel has enabled developers worldwide to dramatically accelerate AI-based development. It’s an open source AI tool kit that makes it easier for developers to write once and deploy AI models anywhere.”
To adapt quickly when processes or raw materials change during mass production, LG Innotek engineers are considering greater reliance on Intel Xeon CPUs—equipped with specialized accelerators—to retrain deep-learning models efficiently. By moving away from setups requiring both CPUs and third-party GPUs toward CPU-only configurations for fine-tuning tasks, the company expects further cost reductions.
