Column

• Time-consuming parameter condition setting for defect feature classification
• Difficulty in parameter condition setting for defect feature classification for those unfamiliar with inspection equipment
• Variability in classification conditions depending on the creator
• Inability to achieve desired classification with current features
• Desire to increase inspection speed

Point 1 : Real-time defect classification in continuous inspection

By having TAKANO AI classify defects simultaneously with detection, high-speed inspection with defect classification becomes possible.

Point 2Feature Feature Classification × TAKANO AI

The technology that combines feature classification and TAKANO AI for defect classification is patented. By integrating the black box portion with conventional feature classification, it also becomes effective for quality control.

Advantages

① Saves time and effort in setting conditions for defect classification

② Capable of capturing high-precision images, thus improving the accuracy of AI-based defect classification

3) Anyone can operate it, regardless of their proficiency with inspection equipment

4) Can be retrofitted to existing equipment

Case: Desire to detect only defects on the glass surface

Feature-based Approach:
Difficult to differentiate with conventional numerical feature classification

Deep Learning Approach:
Deep Learning can classify defects by simply providing "images" and "answers."

Aiding inspection automation with the further possibilities of AI

TAKANO AI enables labor saving in what was previously a time-consuming process of parameter condition setting for feature classification.
However, relying solely on AI for judgment would not lead to true quality control, so we will combine it with conventional feature classification to achieve higher-precision defect classification.
Going forward, we plan to consider adding new features and having AI assist with equipment operation and parameter settings currently performed manually.