Post-Tai Po fire-safety · Hong Kong

Drone-based hyperspectral inspection for fire-retardant safety nets

We verify whether scaffolding safety nets are genuinely flame-retardant — without cutting a single sample. Real-time, full-facade, regulatory-grade chemical verification from the air.

Technology for Safety. Innovation for Humanity.

>95%
Detection accuracy
−70%
Faster than lab sampling
100%
Full-facade coverage
Drone scanning a high-rise scaffolding facade, with a hyperspectral compliance heat-map overlaid on the safety nets
Compliance heat-map Untreated Treated

Backed & recognised by

Cyberport CCMF Grantee HKSTP Ideation Programme 🎓HKU Dean's I&E Award 2025–26 🏆12th HK Student I&E Competition Tam Wing Fan Innovation Fund Philomathia Foundation Fund The University of Hong Kong Cyberport CCMF Grantee HKSTP Ideation Programme 🎓HKU Dean's I&E Award 2025–26 🏆12th HK Student I&E Competition Tam Wing Fan Innovation Fund Philomathia Foundation Fund The University of Hong Kong
The current problem

Today's safety-net checks are manual, slow and only skin-deep

Under Hong Kong's tightened rules, fire-retardant nets must be lab-tested before installation and spot-checked once in use. The process protects the public — but it is destructive, disruptive, and misses most of the facade.

Disruptive & slow

Manual sampling can take days per site, delaying construction schedules.

Spot-checks only

Small cut samples mean non-compliant nets elsewhere on the building are easily missed.

Manual bottlenecks

Reliance on manual labour creates inefficiencies and potential safety gaps.

Mandatory

📋 The regulatory workflow

1
Pre-installation testing

Nets sampled on-site on delivery using a government-prescribed method — fully video-recorded.

2
Government lab approval

Samples sent to a designated lab for flame-retardant testing; nets cannot be installed until approved.

3
In-use surprise sampling

BD and supervisory departments take random samples from installed scaffolding.

4
Enforcement

Non-compliance triggers removal orders, work suspension and significant fines.

The solution

A drone that reads the chemistry of every net

Our drone-based hyperspectral inspection system delivers non-destructive, real-time verification of fire-retardant safety nets across the entire building facade.

✂️

Non-destructive

No need to cut samples — the integrity of the installed safety net is fully preserved.

Real-time analysis

Immediate on-site data processing and treated / untreated classification.

🏢

Full-facade coverage

The drone efficiently scans the entire building exterior, not just sample patches.

🎯

High accuracy

Over 95% precision distinguishing treated from untreated nets.

From take-off to compliance report in five steps

A fully automated workflow built on commercial drone hardware and a purpose-trained spectral AI model.

01

Automated scan

A DJI Matrice 400 with a hyperspectral sensor scans the building facade.

02

Spectral capture

The sensor captures the unique reflected spectral signature of each net.

03

AI analysis

AI model identifies chemical peaks from flame-retardant additives.

04

Real-time class

Netting is classified live as “Treated” (green) or “Untreated” (red).

05

Report generation

An automated PDF compliance report is produced for the Buildings Department.

The platform

Proven hardware

Built on the industry-standard DJI Matrice 400 and a SWIR hyperspectral sensor.

DJI Matrice 400
SWIR hyperspectral sensor
DJI Matrice 400 inspection drone with gimbal-mounted hyperspectral sensor
Fly Fly Check team lab-testing the inspection drone against fire-retardant safety nets
R&D · Lab validationBench-testing the rig against real treated & untreated nets

Ground-truth spectra are collected in the lab to calibrate the on-board classifier before field deployment.

Drone platformDJI Matrice 400
Sensor typeSWIR hyperspectral
Critical bands20–50 contiguous · field-optimal
Operating distance10–50 m · optimal 15–30 m
AI modelOn-board classifier + anomaly detection
OutputLive class + automated PDF report
Commercially available >95% accuracy Non-destructive Adapted from polymer-industry detection
The technology

Seeing the chemistry the eye can't

Flame-retardant additives leave tell-tale fingerprints in the short-wave infrared.

Critical absorption peaks

Our SWIR hyperspectral sensor fully covers every diagnostic peak for fire-retardant detection.

1440 nm  ATH — aluminium trihydrate signature
1580 nm  APP — ammonium polyphosphate signature
1700–1750 nm  Polymer (nylon / PP) baseline reference

How the sensor reads a net

The drone illuminates the material; reflected SWIR waves carry a spectral signature unique to its chemistry.

Drone scanning a high-rise scaffolding facade, with a hyperspectral compliance heat-map overlaid on the safety nets
Emitted & reflected EM waves  build a per-pixel spectral signature
Sensing approachAccuracyDrone fitOptimal distanceKey limitationFeasibility (HK)
Traditional lab100% (lab)Slow, disruptiveImpractical
Visual / thermal<30%Excellent10–100 mNo chemical detectionLow
NIR spectrometer90–95%Good0.5–5 mSlow scanningMedium
Hyperspectral — our choice>95%Excellent10–50 mNeeds calibration / AIHighest
Terahertz (THz)85–93%Moderate1–20 mWeather-sensitiveHigh
mmWave radar<50%Excellent5–100 mPoor chemical sensitivityLow
In the field

From the lab bench to the city skyline

One platform, proven across competitions, community projects and real inspection scenarios.

Proven on real sites

Drone operations, machine-learning R&D and live POCs across Hong Kong.

Social impact

Safer Buildings. Stronger Communities.

Our drone-based hyperspectral inspection system enhances public safety and builds a more resilient Hong Kong.

1
Protects lives and property

Ensures fire-retardant nets perform as intended, reducing fire spread and falling debris.

2
Strengthens public safety

Supports government enforcement with reliable data and greater confidence in construction safety.

3
Faster, smarter, less disruption

Non-destructive, efficient inspections minimise disruption to sites and the surrounding community.

4
Promotes sustainable construction

Encourages consistent use of compliant materials, supporting sustainability and reduced waste.

5
A safer future for Hong Kong

Contributes to a smarter, safer and more resilient city through technology and innovation.

Public Safety

Reduces fire risk and protects lives

−70%

Faster inspections vs. traditional methods

100%

Full-facade coverage for higher compliance

Sustainable

Supports responsible construction & well-being

✦ First-mover advantage

Is this already on the market? No — not yet.

No commercial product offers drone-based hyperspectral inspection specifically for verifying the fire-retardant properties of scaffolding safety nets. Hyperspectral + AI chemical-additive detection for post-Tai Po regulation is novel and first-mover in the Hong Kong / Greater China construction-safety market.

Traditional lab samplingManual testing — the current HK regulatory standard.
General drone inspectionVisual, thermal or multispectral cameras for structural defects — no chemistry.
Net manufacturing & certificationFactory-side certification, not on-site verification via drones.
Fly Fly CheckThe only regulatory-grade, on-site chemical verification for fire-retardant nets.
Awards & news

Recognised, funded and on the radar

2026
Dean's Innovation & Entrepreneurship FundEntrepreneurship Award
May 2026
12th HK University Student I&E CompetitionAwardee
Jun 2026
HKSTP Ideation ProgrammeProgramme grantee
Jul 2025
RTHK news featureDrone service provider for the weather-satellite student project at 保良局蕭漢森小學
Apr–Sep 2025
Tam Wing Fan & Philomathia Innovation FundsInnovation fund awardee
Apr–Sep 2025
Cyberport Creative Micro Fund (CCMF)Grantee
Fly Fly Check team on stage at the 12th Hong Kong University Student Innovation and Entrepreneurship Competition
May 2026 · Competition12th HK University Student I&E Competition

The team representing the University of Hong Kong at the award ceremony.

RTHK news segment featuring the student drone project at 保良局蕭漢森小學
Jul 2025 · RTHK featureDrone service for 保良局蕭漢森小學

Guiding a primary-school weather-satellite student project as drone service provider.

Contact us

Let's make Hong Kong's facades safer

Talk to us about pilots, Buildings-Department integration, or partnering on the next post-Tai Po inspection deployment.

Email the team
Address
Room 1801, 18/F, Hong Kong Plaza,
188 Connaught Road West, Shek Tong Tsui, Hong Kong