189 images of printed circuit boards across 4 defect segmentation classes. The dataset provides pixel-level masks for dry joints, incorrect installations, PCB damage, and short circuits to support automated industrial inspection.
Use Cases
- Train a semantic segmentation model to identify 'short_circuit' regions using the image and mask data.
- Develop an automated optical inspection (AOI) system to flag 'incorrect_installation' defects using the training split.
- Evaluate the precision of defect detection for 'dry_joint' and 'pcb_damage' using the 36-image test split.
Strengths
- 189 images distributed across train (128), valid (25), and test (36) subsets.
- 4 defect classes: dry_joint, incorrect_installation, pcb_damage, and short_circuit.
- Pixel-level segmentation masks for precise defect localization on circuit boards.