INITIALISING DASHBOARD

Loading Hyperspectral AI Models...

Drone-Based Hyperspectral Corrosion Intelligence

Project: PETRONAS Offshore Pilot Brief: Client Technical Review

Client Technical Brief: Drone-Based Hyperspectral Corrosion Classification

Dashboard ini dipadankan dengan technical brief: hyperspectral survey digunakan sebagai inspection layer baharu untuk mengesan surface corrosion indicators, coating degradation, salt deposits, marine growth dan priority zones untuk targeted close-up/NDT inspection. Output ini menyokong keputusan inspection dan maintenance; ia bukan pengganti wall-thickness, CUI atau structural integrity assessment.

Type: VNIR Range: 400-1000 nm Resolution: 2.7 nm Pixels/Line: 900

Resonon Pika L VNIR

400–1000 nm • 281 bands • 2.7 nm

Inspection Gap

Visual-only survey can miss early coating breakdown, salt deposit and oxidation indicators. HSI adds spectral evidence before targeted access work.

Spectral Fingerprint

RGB sees colour; multispectral sees selected bands; hyperspectral creates hundreds of narrow-band reflectance fingerprints for material separation.

Claim Control

Dashboard classifies surface indicators and priority zones. Critical wall-loss, CUI and structural integrity must be validated by NDT and corrosion engineer review.

Pilot Methodology

Client-ready workflow from class definition to corrosion intelligence package.

01

Define Classes

Agree classes, acceptance criteria and report format.

02

Build Library

Coating, rust stages, salt, wetness, marine growth.

03

UAV Acquisition

Pika L capture with calibration and RGB context.

04

Ground Truth

Inspector label, close-up photo, UT / RBI tie-in.

05

ML Validation

SVM, RF, XGBoost, 1D-CNN; F1 and confusion matrix.

06

Deliver Package

Maps, confidence, priority NDT and executive report.

Total Area Inspected

12,450

Corrosion Detected

8.7%

Severe Hotspots

3

NDT Points Req.

22

Asset Corrosion Risk Index

72 High Risk

Severity Breakdown (Area m²)