The performance of Solder Defect Detection:
Research/Service Description:
Apply simultaneous multitask and deep learning techniques to enhance the performance of defect detection. The defect image detection system delivers stable, reliable classification and segmentation results with minimal number of defect images.
Research/Service Description:
Apply simultaneous multitask and deep learning techniques to enhance the performance of defect detection. The defect image detection system delivers stable, reliable classification and segmentation results with minimal number of defect images.
Research/Service Outcome:
A model and/or an Automatic Optical Solder Defect Detection system to identify and classify defects in production line.
 |
 |
 |
| Figure 1. The multi-task learning model for SDD (with mask on the solder patch) |
Figure 2. Solder Defect Detection hardware setup | Figure 3. The SDD detection output with five classes output: normal, no solder, insufficient solder, bridging solder and excessive solder |