A maintenance inspector photographs 847 welds across a chemical plant during a quarterly inspection cycle. Manual review takes 6 days. Fatigue causes a hairline stress-corrosion crack to be rated "acceptable" — it propagates into a through-wall failure 11 months later, costing $740,000 in emergency repair and lost production. OxMaint's AI Vision module analyses those same 847 photos in under 5 minutes at 99% detection accuracy — classifying every defect type, scoring severity on a 1–5 scale, and generating prioritised work orders with annotated images automatically. Human inspectors miss 20–30% of early-stage defects under real conditions, with accuracy degrading a further 20% after two hours of repetitive evaluation. AI vision maintains identical detection performance from image 1 through image 100,000. Book a demo to see OxMaint's AI Vision module running on representative photos from your facility.
Human Visual Inspection
Misses 20–30% of early-stage defects
Accuracy drops 20% after 2 hours
Inter-inspector agreement only 55–70%
Manual work order creation — hours of admin
Cannot detect sub-0.3mm cracks consistently
vs
OxMaint AI Vision
95–99% detection accuracy, every image
Consistent from image 1 through 100,000
Objective defect classification — no inspector variance
Work order auto-generated from detected defect
Detects cracks as small as 0.08mm
What OxMaint AI Vision Detects — And What It Creates
Corrosion
Detects micro-rust and pitting 6–12 weeks before structural compromise. Classifies stage 1 to stage 5 progression on the image.
Cracks & Fractures
91% detection accuracy on cracks under 0.5mm width. Detects cracks as small as 0.08mm on curved surfaces with appropriate imaging setup.
Wear & Seal Degradation
Measures belt edge fraying, surface glazing, and seal lip degradation — quantifying remaining useful life instead of relying on subjective "looks worn" judgement.
Missing Components
Identifies missing fasteners, guards, caps, and labels against a reference template of what the asset should look like — catching safety hazards before an incident.
Leaks & Staining
Identifies oil, coolant, and chemical staining patterns that indicate active or recent leaks — flags source location for targeted work order assignment.
Auto Work Order
Every detected defect triggers an auto-generated work order — severity-ranked, asset-linked, and assigned — in OxMaint. No manual documentation. No defect-to-action gap.
OxMaint AI Vision Module
Smartphone capture → AI defect detection → work order in OxMaint. No separate analytics tool. No manual documentation step.
AI Vision Inspection Performance Data
| Metric | Manual Inspection | OxMaint AI Vision |
|---|---|---|
| Defect detection accuracy | 70–80% under real conditions | 95–99% consistent accuracy |
| Accuracy after 2 hours | Drops 20% — fatigue-driven misses | Unchanged — no cognitive fatigue |
| Minimum detectable crack | 0.3mm on flat surfaces in good light | 0.08mm with appropriate imaging |
| Time to review 847 photos | 6 days including write-up | Under 5 minutes with annotated output |
| Work order from defect | Manual entry — 1 to 4 hours admin | Automatic — generated at detection |
| Inspector agreement on severity | 55–70% — same defect, different verdict | 100% consistent — quantified criteria |
| Audit trail for compliance | Paper or manual spreadsheet — gaps common | Complete digital record linked to asset |
Expert Review
Suresh Pillai — NDT Level III Inspector and Maintenance Technology Consultant, 19 years
AI vision does not make human inspection irrelevant — it makes human inspectors vastly more effective by removing the cognitive burden of reviewing hundreds of similar images looking for subtle anomalies. An experienced NDT inspector's value is in interpreting ambiguous findings, making engineering judgements, and understanding plant-specific context. When AI pre-screens a thousand images and presents the inspector with 12 flagged findings already ranked by severity, the inspector's expertise is applied where it actually matters — not exhausted on routine visual scanning where human error rates are highest. The work order auto-generation piece in OxMaint is what closes the loop — detected and documented in one step, no information lost in handoff between inspection and maintenance planning.
Frequently Asked Questions
Does OxMaint's AI Vision module require specialist camera hardware?
Standard smartphone cameras capture sufficient image quality for the majority of OxMaint AI Vision defect types — corrosion, surface damage, missing components, and visible leaks. For crack detection below 0.2mm or thermal anomaly detection, OxMaint supports integration with affordable macro-lens attachments, industrial inspection cameras, and thermal imaging devices. Book a demo and an OxMaint specialist will recommend the optimal capture setup for your specific defect types.
How does the AI Vision model learn to recognise defects specific to our equipment and materials?
OxMaint's AI Vision model is pre-trained on a large library of industrial defect images across corrosion, cracks, wear, and contamination types. For plant-specific materials or unusual geometries, the model is fine-tuned using a sample of your own labelled defect images — typically 200 to 500 images per defect class. Fine-tuning takes 2 to 4 weeks and significantly improves detection accuracy on your specific asset population. Sign in to OxMaint to upload sample images and initiate model fine-tuning for your facility.
Can the AI Vision module handle routine inspection checklists as well as defect detection?
Yes. OxMaint's AI Vision module supports guided inspection checklists where the technician photographs each asset point in sequence — the AI confirms each photograph against the expected reference state (correct component present, guard in place, label visible) and flags any deviation automatically. This combines checklist completion verification with defect detection in a single capture workflow. Book a demo to see the guided inspection workflow for a representative equipment type.
How does AI Vision documentation support compliance and insurance audits?
Every image captured through OxMaint AI Vision is timestamped, geotagged, linked to the specific asset and work order record, and stored with the AI detection result and severity score as part of the asset's permanent maintenance history. This creates a complete, tamper-evident audit trail for compliance frameworks including ISO 55001, OSHA inspection requirements, and insurance assessment programmes — replacing paper inspection logs that are frequently incomplete or illegible. Sign in to OxMaint to configure audit-ready inspection records for your compliance programme.
OxMaint · AI Vision · Defect Detection · Work Order Documentation
The defect your inspector will miss on image 623 of an 847-image inspection round will never be missed by OxMaint AI Vision. Photograph it. The AI finds it. The work order creates itself.
Corrosion · Cracks · Wear · Missing components · Leaks · Auto work order generation · Full audit trail
