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.
Backed & recognised by
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.
📋 The regulatory workflow
Nets sampled on-site on delivery using a government-prescribed method — fully video-recorded.
Samples sent to a designated lab for flame-retardant testing; nets cannot be installed until approved.
BD and supervisory departments take random samples from installed scaffolding.
Non-compliance triggers removal orders, work suspension and significant fines.
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.
Automated scan
A DJI Matrice 400 with a hyperspectral sensor scans the building facade.
→Spectral capture
The sensor captures the unique reflected spectral signature of each net.
→AI analysis
AI model identifies chemical peaks from flame-retardant additives.
→Real-time class
Netting is classified live as “Treated” (green) or “Untreated” (red).
→Report generation
An automated PDF compliance report is produced for the Buildings Department.
Proven hardware
Built on the industry-standard DJI Matrice 400 and a SWIR hyperspectral sensor.
Ground-truth spectra are collected in the lab to calibrate the on-board classifier before field deployment.
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.
How the sensor reads a net
The drone illuminates the material; reflected SWIR waves carry a spectral signature unique to its chemistry.
| Sensing approach | Accuracy | Drone fit | Optimal distance | Key limitation | Feasibility (HK) |
|---|---|---|---|---|---|
| Traditional lab | 100% (lab) | — | — | Slow, disruptive | Impractical |
| Visual / thermal | <30% | Excellent | 10–100 m | No chemical detection | Low |
| NIR spectrometer | 90–95% | Good | 0.5–5 m | Slow scanning | Medium |
| Hyperspectral — our choice | >95% | Excellent | 10–50 m | Needs calibration / AI | Highest |
| Terahertz (THz) | 85–93% | Moderate | 1–20 m | Weather-sensitive | High |
| mmWave radar | <50% | Excellent | 5–100 m | Poor chemical sensitivity | Low |
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.
Safer Buildings. Stronger Communities.
Our drone-based hyperspectral inspection system enhances public safety and builds a more resilient Hong Kong.
Ensures fire-retardant nets perform as intended, reducing fire spread and falling debris.
Supports government enforcement with reliable data and greater confidence in construction safety.
Non-destructive, efficient inspections minimise disruption to sites and the surrounding community.
Encourages consistent use of compliant materials, supporting sustainability and reduced waste.
Contributes to a smarter, safer and more resilient city through technology and innovation.
Reduces fire risk and protects lives
Faster inspections vs. traditional methods
Full-facade coverage for higher compliance
Supports responsible construction & well-being
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.
Recognised, funded and on the radar
The team representing the University of Hong Kong at the award ceremony.
Guiding a primary-school weather-satellite student project as drone service provider.
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 team188 Connaught Road West, Shek Tong Tsui, Hong Kong