The particle count re evaluation system is used to analyze the detection report data (particle count) of friendly particle indentation inspection machines. For products with particle counts less than the set threshold, traditional image algorithms and AI deep learning systems are used for secondary re evaluation, greatly improving the detection accuracy. The real NG data is uploaded through CIM and locked.
Project |
Performance index |
Testing specifications |
Flexible setting of particle count thresholds for different models and regions. |
For products with particle count less than the set threshold, perform secondary judgment, filter out misjudgments caused by Dimple, dirt, image blur, deviation, etc., and accurately detect genuine NG products. |
Usage effect
1. The misjudgment rate of particle indentation detection machine is generally around 5%. Calculated based on the daily production of 3000 pieces on one production line, the number of misjudgments on one production line is about 150 pieces. The number of manual re judgments required on multiple production lines is even greater. Due to the instability of manual re judgments, it is easy to miss detections, often requiring two rounds of manual re judgments, and there is still a risk of missed detections;
2. After our particle count re judgment system, automatic judgment can be achieved, reducing the number of misjudgments in one line from 150 to around 7.5, greatly improving detection accuracy and reducing manual re judgment workload;
3. Machine detection is stable, reducing misjudgments and missed detections, improving the yield rate of production lines, and reducing the workload of personnel re evaluation. The investment cost can be recovered in three months.
Defect legend