As all lenders have “skin in the game” on every loan, rigorous quality control across the lending lifecycle is imperative. However, manual reviews of appraisals, property condition, renovation estimates, and more do not scale efficiently. Meanwhile, incomplete or inaccurate information can lead to faulty lending decisions that put lenders and their AMCs at risk.
Today, new technology can assist appraisers, AMCs and lenders with visual inspection, analysis, and intelligence to validate information, detect issues, and derive data-backed insights on every loan. It can also accelerate the review process. It is not a tool to replace appraisers. This new, vital tech tool – computer vision – is how appraisers can better streamline their workflow, while increasing productivity and accuracy.
What is computer vision?
Computer vision is a technology that enables computers to interpret and understand visual information from around the world, akin to human sight, but with greater precision and consistency. It utilizes the latest artificial intelligence and deep learning innovations to extract information from visual data.
Computer vision can identify, categorize, and comprehend details in images and videos as much as the human eyes and brain can discern objects, textures, colors, and more – only better and faster. Computer vision doesn’t get tired and catches things human eyes often miss.
Over the past decade, computer vision capabilities have advanced enormously thanks to exponential data and processing power growth. Processing speeds have increased by a factor of 10 since 2015, to 1.9 exaflops today on a high-performance system. The cost to store 1 GB of data has decreased from nearly $10,000 in 1990 to less than $0.01 today.
According to Epoch research, “Since 2010, the amount of training computed for machine learning models has grown by a factor of 10 billion, significantly exceeding a naive extrapolation of Moore’s Law.” As a result, computer vision models and use applications continue advancing at remarkable speed and increased efficiency.
In the lending arena, computer vision tools can consume photos, video, and other visual collateral data to automate inspection for damage, measure rooms, detect defects, identify property styles and features, assess conditions, create 3D models, and much more. As computers progress rapidly, they are beginning to bring unprecedented levels of automation, analytics, and precision by unlocking data to increase risk assessment and quality control in mortgage lending.
The crucial role of appraisers
In the residential financing industry, appraisers are lending’s unsung heroes, providing the human power to assure that banking institutions do not falter. But being an appraiser today brings risks and rewards, making the accuracy of property valuations and quality control not just a priority but a necessity. It’s why the most successful appraisers are often at the forefront of adopting best practices for quality control and due diligence.
New computer vision tools built specifically for the mortgage industry are used to determine property condition and quality scores, create damage models, find and evaluate comparable data sets, and more. Most importantly, lenders, along with Fannie Mae and Freddie Mac, are leveraging computer vision to both scale and streamline their loan quality and condition review processes – two things that are typically in conflict with each other.
By leveraging the power of computer vision, they are developing scalable processes that incorporate best practices and objective third-party data to arrive at more accurate valuations. They also are using computer vision to validate third-party valuations prepared by appraisal management companies.
Why lenders are now rapidly deploying computer vision
Most recently, computer vision has been on fire in the mortgage lending world as lenders seek to align with new GSE guidelines and practices. This year, the GSEs made a significant shift by evaluating photos for condition and rejecting appraisals when comparables are used without the appropriate adjustments.
Why? Fannie found the power of computer vision to be incredibly accurate. For example, Fannie Mae analyzed more than a million appraisals using computer vision or image recognition technology to compare interior photos of comparables for condition ratings. The ratings were so significantly different that Fannie Mae now uses computer vision technology to improve its conditions ratings process.
Fannie Mae’s use of computer vision technology identified appraisal defects with a 98% accuracy rate and significantly increased its appraisal condition review efficiency. Fannie also noted that computer vision allows it to find “many defects that were previously impossible” for it to detect.
Fannie Mae’s innovative application of computer vision is dramatically improving its review processes. By analyzing over a million appraisals, Fannie Mae enhanced its condition ratings and identified defects with near-perfect accuracy, a previously unimaginable feat.
Transforming quality control and collateral due diligence
The heart of every lender’s operation resides in the accuracy of property valuations. Here’s where computer vision proves its mettle.
It’s a system that can analyze hundreds of property images, comparing and contrasting them against a vast database, providing insights that are humanly impossible to discern. This level of detail and precision in assessing property conditions and values can significantly reduce the risk of overvalued collateral, which is often challenging for lenders.
For appraisers, computer vision is like having an eagle-eyed assistant on your team, but one that works tirelessly, without error, never calling in sick or taking a single vacation day.
Balancing scale with quality
For lenders, scaling up loan review processes typically means increasing risk by compromising quality: it becomes a tug-of-war between expanding a portfolio and maintaining the quality of loan reviews.
Computer vision offers a harmonious solution to lenders too. By automating and enhancing the accuracy of property evaluations, they can grow their portfolios without sacrificing the thoroughness and reliability of their loan review processes.
Revolutionizing ‘Fix and Flip’ lending
In the ‘fix and flip’ sector, accurate valuation of properties before and after renovation is crucial. Traditionally, this process has been labor-intensive, reliant on manual searches and subjective judgments.
Here, computer vision introduces a paradigm shift. It provides comprehensive data on property conditions, enabling precise before-and-after evaluations. Computer vision eradicates this uncertainty, offering detailed data on property conditions of comparables and neighborhood market trends. This not only helps appraisers streamline their processes but also enhances the reliability of their work.
Gaining deeper market insights and trends
Successful appraisers know that having a deep understanding of local market trends is essential. Here, computer vision extends its utility beyond mere valuations. It can analyze vast amounts of visual data and use that data to extrapolate additional information from other data sources.
The results generated through this cutting-edge technology can offer insights into trending home features, popular designs, and even regional preferences. This data can be invaluable for appraisers, ensuring that renovations and improvements align with market demands. It’s like having a multi-tool that assists in one aspect of your business and simultaneously enhances other facets.
Embracing future technology today
Integrating computer vision into your processes is not just about keeping up with technological advancements; it’s about leading the charge in the Modernization of the appraisal industry. Adopting effective, proven, leading-edge technology gives you a strategic advantage as an appraiser, setting you apart from others.
Navigating your path forward
The journey to incorporate computer vision into your appraisal practices may seem daunting, but the potential benefits can be immense. The key to the successful adoption of computer vision is to start small and start using it to understand the capabilities of this new technology and then gradually integrate it throughout your existing processes. Once you see the benefits in action, it’s time to leverage all the ways computer vision can help you.
Looking ahead
Embracing computer vision is like a master navigator charting a course through unexplored waters with the aid of advanced maritime instruments. Just as these instruments enable the navigator to steer clear of hidden reefs and find the most efficient routes, computer vision guides you to better navigate the complex seas of property valuation and the impact your work has on loan quality control.
Computer vision can help you avoid the pitfalls of inaccurate appraisals and help provide more reliable data and improve decision-making. With computer vision as your compass and sextant, you are set to ensure that your journey is safer and more successful.
Share this article
Written by : Tony Pistilli
As President of Valuations at Restb, Tony Pistilli is responsible for providing direction to the application of Restb.ai's products and services for the valuation segment of the real estate industry, working with the product team to develop and expand the suite of offerings and prioritizing development initiatives. Tony also plays a vital role in expanding Restb.ai's reach in the valuation and appraisal industry, as well as fostering relationships with lenders and related industry partners.
Tony has over 30 years of executive-level real estate valuation and lending experience including working with national banks, mortgage companies, federal agencies, and leading appraisal management firms. He is a certified residential real estate appraiser in Texas and is an AQB Certified USPAP Instructor. In 2011, he was the first recipient of the Valuation Visionary Award presented by the Collateral Risk Network at Valuation Expo.