AI deep learning vision software company Neurocle has entered a strategic partnership with Taiwan-based global hardware specialist Portwell to deliver edge-based AI vision inspection solutions optimized for diverse industrial hardware environments.

Portwell, headquartered in Taiwan, is a leading provider of industrial computing and edge AI systems. The company offers a broad range of industrial platforms, including high-performance embedded systems based on x86 and ARM architectures, as well as solutions for medical applications, factory automation, network security, and AI edge devices. Portwell has established strong capabilities in NVIDIA Jetson-based edge AI hardware platforms, recently integrating high-efficiency AI accelerators to achieve stable, high-speed inference performance even in compact form factors.

Neurocle develops Auto Deep Learning-based vision inspection software, Neuro-T and Neuro-R, enabling users to build high-performance inspection models without requiring specialized deep learning expertise. The company provides a total of nine AI models, covering functionalities such as defect generation, patch-based classification, and unsupervised anomaly detection. These solutions are applied across a wide range of industries, including semiconductors, food and beverage, automotive, steel, batteries, and biotechnology, supporting quality inspection processes.

With Auto Deep Learning algorithms that automatically optimize model architectures and hyperparameters, Neurocle delivers high-performance models that combine both accuracy and speed. Its solutions have received strong market recognition for their ability to address real-world challenges in industrial environments.

Through this partnership, the two companies will jointly offer an integrated solution capable of performing high-precision vision inspection in real time even in edge device environments. By applying Neurocle’s deep learning model optimization to Portwell’s NVIDIA Jetson-based PJAI series hardware, the solution achieves inference speeds approximately 2.5 times faster than comparable edge platforms with similar specifications.

This advancement enables real-time delivery of high-precision inspection results and allows for the deployment of stable, lightweight AI inspection systems without the need for separate servers. The solution is expected to maintain high performance even under constrained computational resources, while significantly improving efficiency in system integration and maintenance.

The first outcome of this collaboration will be unveiled at BUTECH 2025 in Busan this May, as well as at COMPUTEX TAIPEI in Taiwan. Live demonstrations will showcase Neurocle’s AI models running in real time on Portwell’s PJAI series hardware, offering an opportunity to view how deep learning models detect real industrial defects in edge environments.

Hongsuk Lee, CEO of Neurocle, stated, “Our collaboration with Portwell will serve as a turning point in accelerating the adoption of edge AI-based vision inspection systems across industries. We plan to continue expanding our technological partnerships to deliver advanced edge AI solutions tailored to diverse industrial environments.”