Selected work · 04

CG-BI: Confidence-gated triage for blade inspection

Routing, not tier choice, is the lever: a two-tier cascade beats both of its own models on quality and lands between them on cost.

MultimodalPythoncost control RepositoryFull report
92.3%critical-defect recall
47%of the strong tier's cost
25.3%exact match, best of any tier
0.424class F1

What it is

A drone captures hundreds of blade images per turbine, and an engineer still writes a finding for each one: defect type, severity, location, action. This measures whether a multimodal model can draft that finding, and what a two-tier routing policy buys over picking a single model.

How it works

  • 75 real drone images across a fixed six-class taxonomy, one call per image unless escalated.
  • A confidence gate. A cheap multimodal model labels every image; anything below 0.9 confidence, 36% of the set, goes to a stronger model.
  • Costed per thousand images, with p50 latency, so a quality win that nobody can afford shows up as one.

What it measured

TierExact setClass F1Critical recallCost / 1,000
Cheap model alone17.3%0.32350.0%$0.43
Strong model alone14.7%0.41796.2%$3.64
Routed cascade25.3%0.42492.3%$1.73

The cascade beats both single tiers on exact match and F1, holds 92% critical-defect recall, and costs 47% of the strong tier alone. The strong tier is accurate but expensive; the cheap tier alone misses half the critical defects.

What it does not claim

One seed, 75 images, and no healthy-blade class, so this is not a claim that inspection can be automated. The cascade drafts a finding for an engineer to approve or correct.