5 Ways Computer Vision is Revolutionizing the Real Estate Industry

Have you been hearing the term “computer vision” a lot lately and wondering what the buzz is all about? In this post, we’ll explain what it is—and how it’s transforming key facets of the real estate sector.

Computer Vision: What Is It?

Computer vision is a form of artificial intelligence (AI) and machine learning that enables computers to derive meaningful information from images and automate actions based on that information—rapidly and at scale.

Computer vision (CV) algorithms trained by data scientists and industry-specific experts help self-driving cars navigate safely from point A to B; spot tumors invisible to the human eye in MRI scans, analyze soil-water balance to optimize crop irrigation, and much more. It’s the same technology that powers Amazon Go stores, which enable customers to pick up items and simply walk out the door, with the transactions tied to their mobile phones—no cash, credit, checks or standing in line required.

According to Allied Market Research, the computer vision market could top $41 billion by 2030—up from just $9.4 billion in 2020.

Today, these same AI-based technologies are gaining traction across a wide spectrum of use cases in the real estate industry. Consumer and business demand for instant, accurate property quotes, the rise of iBuyers, growing interest in single family residential (SFR) investing, and increased competition among institutional investors and traders are just a few of the factors driving computer vision adoption in the pursuit of fast, accurate property intelligence.

Here are just a few examples of how computer vision is transforming operations throughout the real estate value chain.

Appraisals On-Demand: Instant, Accurate Valuations

Amid a shortage of appraisers, ongoing COVID distancing measures, and longer closing periods, investors, lenders, underwriters, and online portals are turning to computer vision technologies to access accurate property condition data.

Today’s most robust solutions use computer vision to analyze high-resolution aerial photography to deliver property condition data on tens of millions of properties available instantly through APIs. Consumer-facing online real estate portals such as Zillow, Trulia, and Redfin factor metadata from this and other forms of image data into their machine learning models to generate home price estimates. In Zillow’s case, computer vision techniques help deliver Zestimates that are within 2% of a home’s actual selling price, instantly and on-demand.

When paired with automated valuation models (AVMs), computer vision automation from objective sources can dramatically enhance the utility and accuracy of desktop property intelligence tools used by lenders, iBuyers, and SFR firms, as well. In real-world implementations, for instance, AVMs augmented with CAPE’s own CV-based property data have shown a 7.7% improvement in PPE10 predictions of on-market valuations.

Beyond Curb Appeal: Visual Property Search

With computer vision, interior and exterior images in MLS listings and other sources can be identified and classified—which is a kitchen, a living room, a primary suite bathroom, etc.—and rated by measures such as modern, dated, average, etc.

Zillow, for instance, uses computer vision to tag and rate objects such as granite counters, island layouts, wall ovens and other details. In addition to improving valuation models, this can also enhance visual property search. Brokerages and others can do the same thing using a new generation of apps such as Foxy ai and Restb.ai.

Instead of searches or recommendations based solely on quantitative data—4-bed, 2-bath, etc.—they can also be qualitative—such as homes in this neighborhood or zip code with updated bathrooms; location of doors and windows relative to park views or morning sun; kitchens with similar amenities, and so on.

Picturing the Possibilities: Estimating Renovation Costs

Metadata captured by computer vision can also price home renovations. Hosta.ai, for instance, uses CV, machine learning, and math to generate measurements, create elevations, and itemize required materials from photos uploaded by mobile phone.

This can be crucial to mortgage lenders, iBuyers, and other flippers in choosing properties, property managers factoring condition data into acquisition or divestiture decisions, and insurers assessing claims for things like water damage.

Even everyday homeowners can use these technologies to price out renovations. Plunk, for instance, is a mobile app that uses CV and image analysis to assess a home’s current value in real time, and recommend specified remodel projects that can translate into a higher listing price.

Virtually Amazing: ‘Digital Twins’ for 3D Tours

Self-guided tours have gained currency during the pandemic, and have changed the way buyers visit single- and multi-family homes, as well as office spaces for rent or purchase.

Matterport, for instance, uses computer vision to scan property interiors in order to render 3D, digital twins for immersive, virtual walkthroughs available 24/7. Its technology automatically blurs out faces and personal information, and enables teams to embed pop-up notes (or “Mattertags”), links, video, and ecommerce workflows into the 3D models.

According to Redfin, Matterport helped it increase monthly virtual walkthroughs by over 600% since the start of the pandemic, and it’s easy to see how everyone from everyday homebuyers to large investment firms making high volume purchases would appreciate the ability to tour homes direct from their desktop—and soon, even via the Metaverse. Firms like OpenSpace and Buildots are bringing these same CV, AI and data visualization technologies to construction sites, as well.

Eagle AI: Fraud Detection

Computer vision has become key to fraud detection in a number of real estate-related industries, including property insurance. Increasingly, restb.ai and others are also using this technology to weed out duplicate listings from real estate portals.

Whether added by accident, by fraudsters, or rival brokerages or search portals, duplicates can diminish the value of a sales or rental portal’s listings. But tasking employees with taking down duplicate listings is costly and time-consuming, while leaving them active will quickly frustrate users.

Computer vision can scan millions of listings and images and remove duplicates quickly, protecting the user experience—and the portal’s integrity. This same technology can also be used to detect mortgage fraud, by comparing borrowers’ income details with bank statements and previous data to alert lenders to validate applications before processing loans.

To learn more about computer vision technology and how it works, check out our post on Geospatial Analytics.

Aggregate Statistics Created Using Data Produced from Nearmap Imagery