Intel and LG Innotek Advance AI-Driven Smart Manufacturing

• LG Innotek adopts Intel’s CPU, GPU and AI tools to build scalable, cost-efficient smart factories with automated inspection and retraining capabilities.

AI-Powered Inspection Enhances Manufacturing Precision

LG Innotek has partnered with Intel to implement an AI-driven inspection system at its Gumi, Korea facility. The collaboration integrates Intel Core and Xeon processors with Arc GPUs to streamline defect detection across production lines. OpenVINO, Intel’s open-source AI toolkit, serves as the backbone for deploying models efficiently across hardware platforms. This setup aims to reduce manual oversight while maintaining high standards in component manufacturing.

The system processes real-time data from the production line using Intel Core CPUs, which handle basic analysis tasks through built-in graphics. More demanding operations, such as running multiple algorithms on high-resolution images, are offloaded to Intel Arc discrete GPUs. Accumulated datasets are transferred to Xeon-powered servers for pre-training and refinement. Future plans include incorporating Intel Gaudi AI accelerators to further optimize training workloads.

Scaling Smart Factories Through Cost Efficiency

LG Innotek’s decision to use Intel Arc GPUs has led to notable cost savings compared to alternative hardware solutions. These efficiencies have enabled the company to scale its AI inspection systems beyond initial deployments. The first implementation took place on the mobile camera module production line, where results demonstrated the viability of the approach. Expansion is now underway at domestic sites like the Gumi 4 factory, which produces flip-chip ball grid arrays, and will continue across international facilities.

The broader goal is to establish a smart factory model that leverages AI to improve competitiveness. By automating inspection and reducing reliance on manual checks, LG Innotek expects to enhance throughput and consistency. Intel’s hardware ecosystem provides the flexibility to adapt to different production environments. As the rollout progresses, the company anticipates further gains in operational efficiency.

OpenVINO and Xeon CPUs Support AI Development

Initial concerns about integrating built-in GPUs were addressed through OpenVINO’s compatibility and ease of use. Engineers at LG Innotek found that the toolkit simplified deployment across various hardware configurations without extensive code rewrites. Since its release in 2018, OpenVINO has helped developers accelerate AI model deployment across edge and server environments. Its role in this collaboration underscores the importance of adaptable software in industrial AI applications.

To support retraining of deep-learning models, LG Innotek is evaluating Intel Xeon CPUs equipped with AI accelerators. These processors offer parallel computation capabilities and support Intel Advanced Matrix Extensions, which enhance training and inference performance. Moving away from third-party GPUs, the company plans to rely more heavily on Xeon CPUs for fine-tuning tasks. This shift is expected to reduce costs while maintaining computational efficiency.