AI · Computer Vision Inspection
Quality Intelligence
32 inline cameras · 14 defect classes · 50ms inference per part with continuous model monitoring.
FPY 96.1%
Inspections / hr
8,420
Defect Rate
0.42%
AI Precision
98.2%
False Positives
0.18%
Live Inspection Feed · Line B · Camera #14
Real-time defect overlay · 50ms inference
REC · 14:42:08
CAM-14 · 4K · 60fps
Model: DefectNet-v4 · TensorRTThroughput 142 parts/min
Defect Taxonomy
Last 24h · 412 defects
Edge chip142
Surface scratch98
Color drift84
Misalignment56
Other32
Defect Stream
Auto-classified · pending review
| ID | Type | Part | Confidence | Line | Time | |
|---|---|---|---|---|---|---|
| DF-9912 | Edge chip | Cabinet shutter Oak | 97.0% | Line B | 12s ago | |
| DF-9911 | Surface scratch | Drawer front Walnut | 92.0% | Line B | 1m ago | |
| DF-9910 | Color mismatch | Side panel 720 | 89.0% | Line C | 2m ago | |
| DF-9909 | Drill misalign | Hinged door 18mm | 95.0% | Line A | 4m ago | |
| DF-9908 | Edge chip | Shelf 600 | 84.0% | Line D | 5m ago |
Model Monitoring
DefectNet-v4 · drift & calibration
Precision98.2%
Recall96.7%
F1 Score0.974
Data driftstable
Last retrain4 days ago
Active versionv4.2.1
FPY Trend · 30 days
First Pass Yield
Defects by Line · today
Hourly distribution