CNAI Improves False Alarm Rate with Synthetic Data Technology

Inspection Equipment False Alarm Rate Improved by 13.3 Percentage Points

CN.AI utilized generative artificial intelligence (AI) technology to improve the accuracy of product inspection equipment, enhancing its potential applications in the manufacturing sector. CN.AI recently announced on the 25th that it conducted a defective product synthetic data generation project with HL Mando to improve the accuracy of product appearance inspection equipment, confirming a reduction in the false detection rate.

This project was carried out to improve the performance of inspection equipment for defective products and strengthen the quality control process. The defective product synthetic data used was created using CN.AI's image generation AI platform, Tivv.

Based on the basic dataset provided by HL Mando, CN.AI utilized generative AI technology to create 7,000 defective product images and trained HL Mando's AI classification model with them. As a result, the false detection rate improved by 13.3 percentage points compared to before.

Synthetic data is virtual data generated according to the statistical distribution of real data and is used to address data scarcity and bias issues, thereby enhancing AI performance. Kim Bohyung, CEO of CN.AI, stated, "Typically, in manufacturing, there are fewer than 100 defective product images available to train AI-equipped product inspection equipment, making it difficult to secure model performance. Through this project, we confirmed the validity and applicability of defective product synthetic data, and we expect the value of synthetic data utilization to increase across various manufacturing fields in the future."

This content was produced with the assistance of AI translation services.