Automating visual inspection with AI achieves both quality and workload
AI SOLUTIONS
Somic Transformation
外観検査の自動化
中小企業
Automating the inspection process on manufacturing lines, where visual inspection is still the norm, with image classification AI for visual inspection. Implementing AI in small and medium-sized enterprises improves operational efficiency and alleviates labor shortages.
課題・ニーズ
- We want to achieve highly accurate anomaly detection even with limited anomaly data.
- We aim to reduce labor shortages, balance workloads, and improve inspection efficiency.
- While high-precision testing is required, we want to reduce the high costs of introducing AI
開発内容
Aiming to automate visual inspections using AI, NABLAS carefully articulates hidden requirements based on the "intuition and tricks" of on-site personnel. NABLAS's AI engineers are stationed at the factory, working closely with the on-site personnel to repeatedly clarify inspection standards and improve the system. Furthermore, rather than aiming for quality that exceeds the judgment of experienced inspectors, the AI is designed to learn and become proficient through actual operation.
成果
- Converts inspectors’ tacit knowledge into AI-readable criteria.
- Using unsupervised models, AI learns abnormal patterns from data on good products and a small number of defective ones, improving detection accuracy.
- Low-cost for SMEs and scalable across industries.
