UC-09 · Scratch · dent · contamination

Surface Defect Detection

Scratches that human eyes miss after the third hour of inspection. Dents on stamped parts, contamination in injection-molded plastics, micro-cracks in glass. AI vision sees consistently — every part, every shift, every defect class flagged with the right severity score.

REC INSPXION X-12B LIVE INSPXION X-12B · SURFACE DEFECT 240 PPM 0.8% MISS LIVE InspXion SPEC RESOLUTION 12 MP LIGHTING POLARIZED CLASSES 4 sev. REPEAT 0.05 mm REJECT · SEV · 1 PARTS 5,128 CRITICAL 1 MAJOR 2
Why it matters

Surface defects are the hardest thing for human inspectors to catch consistently.

Fatigue, lighting variation, and the sheer volume of parts mean even great inspectors miss 5-15% of real defects. Worse, they over-reject borderline parts because 'when in doubt, throw it out.' AI inspection holds the line at the same severity threshold every time, with image evidence for every reject.

Two paths to the same outcome

Retrofit, or build it in from the start.

Retrofit · use what you have

If you do manual visual inspection.

Most lines have an inspection station with bright lighting and an inspector. We add a camera at the same station; the inspector keeps doing what they do for the first 4 weeks while the AI learns from their decisions. After training, the AI runs autonomously; the inspector becomes the exception handler — only the borderline cases come to them.

  • Add to existing inspection station; no rebuild required
  • Co-pilot mode for first 2-4 weeks of training
  • Inspector becomes exception handler, not 100% inspector
  • Severity scoring: critical · major · minor · cosmetic
  • Image archive of every reject for customer disputes
Greenfield · spec from scratch

If you're designing a new inspection station.

Spec the IX-AS/01 with project-tuned lighting at design time. Polarized for glossy metal, dome for textured plastic, coaxial for highly reflective. The camera + lighting + AI station becomes the primary inspection — fully automated, with an exception-handling station for borderline calls.

  • Lighting matched to substrate: polarized, dome, ring, coaxial
  • Per-defect-class detection with severity scoring
  • 100% inspection at line speed (no sampling)
  • Reject diversion via dry contact or PLC
  • Defect telemetry: per-shift, per-mold, per-tool tracking
How it works

Four steps from camera to verdict.

01

Light

Project-specific lighting illuminates the part to maximize defect contrast.

02

Capture

Multi-pose imaging if needed — top, side, angled. 12 MP per pose.

03

Classify

AI detects defects by class and severity. Borderline cases flagged for human review.

04

Sort

Pass / minor / major / critical verdicts to PLC. Reject sortation by severity class.

Compatible with

The InspXion stack for this use case.

Ready to put this into production?

Get a sized configuration in 48 hours.

Tell us your line, your existing equipment, and your tolerance window. We'll come back with a retrofit assessment, a greenfield specification, and a rough cost estimate within two business days.