How Computer Vision Is Replacing Property Inspections

Published on offmarket.now | March 2025

A property inspection used to mean a clipboard, a torch, and a building inspector crawling through a sub-floor. Increasingly, it means a camera — mounted on a drone, a satellite, or the buyer's own phone — and software that can identify defects humans miss.

What's changed in the last year

Moody's acquired CAPE Analytics in January 2025, merging catastrophe risk modelling with geospatial computer vision. CAPE's technology covers 110 million structures using aerial and satellite imagery, automatically assessing roof condition, vegetation encroachment, and structural characteristics. Their Roof Condition Rating is already used by more than half of the top US insurance carriers.

The acquisition signals something important: property condition assessment is moving from periodic human inspection to continuous automated monitoring. If an insurer can assess your roof from a satellite image, a buyer can assess an off-market property without ever stepping on the lot.

Construction-focused computer vision has also advanced rapidly. Models trained on thousands of annotated defect images now achieve 90%+ accuracy in identifying moisture intrusion, warping, peeling paint, and early-stage mould — all from photographs. What previously required a week-long site inspection can now be triaged in hours from photo sets.

The industry is moving toward what some firms call "digital twins" — continuously updated 3D models of a property maintained through regular interior scans. When a property changes hands, the digital twin provides a condition baseline far more detailed than any traditional inspection report.

Why this matters for off-market buyers

Off-market transactions compress timelines. A buyer who learns about a property on Monday might need to make a decision by Friday. There often isn't time for a full building inspection, or the vendor may resist giving access for one.

Computer vision tools help in several ways:

• Remote pre-screening — aerial imagery can reveal roof condition, drainage issues, vegetation risk, and structural anomalies before you've even driven past
• Photo-based triage — if the agent provides interior photos (common in off-market), AI can flag potential issues: staining patterns consistent with water damage, cracks in visible foundation sections, electrical panel age indicators
• Comparative assessment — by analysing thousands of similar properties, models can estimate the likely condition of components based on age, material, and location. A 30-year-old terracotta roof in a coastal suburb has a predictable degradation profile
• Insurance risk scoring — the same CAPE-style models that insurers use can inform a buyer's due diligence. If the insurance industry considers a property high-risk based on imagery, that's a signal worth understanding before purchase

What it can't do

Computer vision is a screening tool, not a replacement for a thorough building inspection. It can't see behind walls, assess plumbing integrity, or test electrical systems. The technology is best understood as a first filter: it identifies properties that warrant closer inspection and flags issues that a human inspector should prioritise.

For off-market buyers operating under time pressure, this first filter is valuable. It's the difference between making a blind offer and making an informed one.

The Australian context

Australia's building inspection industry is largely unregulated in most states — there's no national licensing standard for building inspectors. This makes AI-assisted assessment both more valuable (as a check on inconsistent human inspection quality) and more necessary.

Australian-specific considerations for computer vision assessment include:

• Termite damage — computer vision can identify termite mud tubes and frass in exterior imagery, though interior detection still requires physical inspection
• Asbestos-era construction — properties built between 1940-1990 are flagged automatically based on construction era and material appearance
• Stormwater and drainage — aerial imagery combined with topographic data can identify properties with poor drainage grades

What to do with this information

If you're evaluating an off-market property with limited time:

• Check aerial imagery — Google Earth, Nearmap, or council aerials can reveal roof condition, extensions (potentially unapproved), and drainage patterns
• Request photos — ask the agent for comprehensive interior photos, not just marketing shots. Look for staining, cracking, and unevenness
• Check insurance assessments — some insurers now provide address-specific risk reports that incorporate aerial condition data
• Still get an inspection — use AI-assisted screening to prioritise which properties are worth the $500-$800 inspection fee, not to skip it entirely

The trajectory is clear: property condition assessment is becoming a data problem, not just a physical one. For off-market buyers who need to move fast and make decisions with incomplete information, that's a meaningful advantage.

---

Browse off-market listings or explore by suburb on offmarket.now.