← Case Studies · Energy & Oil and Gas

10× inspection throughput.
API 1104 compliant.

A Vision Intelligence deployment. AI-assisted radiographic weld defect detection across pipeline kilometres.

YOLOv8Edge GPURadiographer consoleInspection report engine

CLIENT CONTEXT

A major pipeline operator with hundreds of kilometres of new and in-service welds requiring radiographic inspection. Qualified radiographers in short supply. Inspection backlogs delaying commissioning.

THE PROBLEM

Radiographic interpretation is skilled, slow work. Every film requires expert review. Volumes were outpacing radiographer capacity. Delays were pushing commissioning milestones and incurring contract penalties.

How We Approached It

Four phases.
Each one auditable.

01

Model training

YOLOv8-based detection model trained on thousands of radiographic films labelled by senior radiographers. Defect taxonomy aligned with API 1104.

02

Confidence-gated triage

Films analysed at the edge. Clean films auto-pass. Ambiguous films flagged with region-of-interest highlights for radiographer review.

03

Radiographer workflow

Flagged films presented with AI annotations. Radiographer confirms or overrides. Every decision logged to the inspection record.

04

Compliance documentation

API 1104 and ASME-compliant inspection reports generated automatically from the decision log.

10× inspection throughput.API 1104 compliant.

Four phases.Each one auditable.

Your case study next.

10× inspection throughput.
API 1104 compliant.

Four phases.
Each one auditable.