Neurocle, an AI vision inspection software company, has achieved the distinction of winning the ‘Innovators Awards’ for five consecutive years, presented by Vision Systems Design, the world’s leading media outlet specializing in machine vision. This year’s award recognizes both the technological innovation and practical industrial applicability of the deep learning models and algorithms Neurocle developed for high-resolution image inspection processes.

In recent years, industries such as semiconductors, precision electronic components, and batteries have increasingly adopted inspection processes utilizing 4K and higher-resolution images, driven by product advancement and growing demand for detecting small and subtle defects. However, high-resolution images often reach hundreds of megabytes in size and require substantial computational resources, making it difficult for conventional AI vision systems to simultaneously achieve real-time processing and high-precision detection.

To address these challenges in industrial environments, Neurocle has advanced a range of Patch-based algorithm technologies specialized for processing high-resolution images. Notably, its Patch Classification model processes entire original images by dividing them into patch units without resizing, effectively mitigating the frequent issue of missed micro-defects that commonly occurs in traditional classification models.

In addition, the Patch Mode of its Segmentation model enables precise detection of micro-defects at the scale of tens of micrometers within specific regions of high-resolution images. This allows for accurate identification of extremely small defects, including electrode coating imperfections, surface scratches on films, and fine cracks.

The Patch Mode applied to the Synthetic Defect Generator model enables the creation of high-quality training data that accurately reflects the size, shape, and location of real micro-defects. This technology allows AI models to learn and recognize the characteristics of ultra-fine defects with greater precision. As a result, it ensures highly reproducible deep learning performance even in high-resolution environments and supports the rapid and stable development of models in new production processes where initial training data may be limited.

Neurocle’s Patch-based technology has effectively brought high-resolution image inspection processes, which were previously difficult to manage with conventional AI systems, into practical industrial use. Even with large 4K and higher-resolution images reaching hundreds of megabytes in size, the technology enables high-accuracy detection of micro-defects, ensuring stable quality control without overlooking extremely small flaws. By delivering technological innovations that directly address the practical needs of industrial sites, Neurocle has once again been selected as a recipient of the Innovators Awards, presented by Vision Systems Design, achieving the milestone of five consecutive wins.

Hongsuk Lee, CEO of Neurocle, stated, “This award is the result of our continuous efforts to advance AI inspection technologies that are truly required in industrial processes,” adding, “We will continue to focus on developing industry-specialized AI technologies capable of proactively addressing challenges in manufacturing environments.”