PAVE by Discovery Loft: Augmenting Industry Workflows with Robust Inspection Results While Bridging Vehicle Data Gap

PAVE by Discovery Loft is making waves in the automotive industry by using computer vision to automate the detailed capture, inspection, and grading of a vehicle enabling anyone to deliver accurate, comprehensive vehicle inspections within minutes.

Since its inception in 2017, Discovery Loft's mission was to take an industry-focus, collaborative approach to technology by working directly with its intended clients to find ways to accelerate the much needed digital transformation for the automotive industry. The company's partners have hands-on involvement in all stages of product development, from very early on, including ideation to going to market and continuous improvements. With this, Discovery Loft discovered PAVE and how to develop it as a product that addresses a real need versus trying to invent a need. This approach also helped the company find the right market fit of how PAVE should interact with the end-users and integrate into the whole industry.

"In partnership with industry innovators Discovery Loft, this new AI-enabled offering continues the digital-first evolution of the Cox ecosystem and enables real-time benefit," said Stephanie Turner, Director, Product and Corporate Strategy, Cox Automotive Canada.

Leaderships with Complementary Skill Sets

Discovery Loft's Co-founders and Co-CEOs Stephen Southin and Brian Steinhauser share rare complementary skill sets. Stephen and his 25+ years of automotive retail and wholesale experience delivers in-depth domain knowledge that was essential in his focus as PAVE's creator and product architect. He also has 15+ years of technical and startup expertise that he gained as an AutoTech entrepreneur; with his second recent successful exit being Bumper App, he brought to market in 2012 and successfully exited through an acquisition by Vicimus Inc. in 2017.

On the other hand, Brian brings decades of executive sales and brand strategy background having experience working with several Fortune 500 brands, including TD Bank Group, Samsung Electronics, Toyota and Magna International. He has a passion for building and maintaining strong enterprise-level relationships with attention to detail that ensures client success by meeting their expectations. Their ongoing mentorship has fused the dual that Stephen and Brian receive from their Partner and Advisor, Ed Clark.

Ed is currently the Chair of the Vector Institute for Artificial Intelligence in Toronto. Previously, Ed served as Group President and Chief Executive Officer of TD Bank Group until his retirement in 2014. Twice during his tenure, Barron's Magazine recognized him as one of the world's 30 Best CEOs. In 2016, Ed was inducted into the Canadian Business Hall of Fame. He is an Officer of the Order of Canada, one of the country's highest distinctions.

Bringing Computer Vision to Customers' Driveway

When trying to buy a car digitally, as people have a vehicle to trade-in, (like over 50% of all online shoppers), the current tools for making a selection for their next vehicle and where to buy it, has come a long way. Still, if they have a car to dispose of as part of their deal, things get tricky. Yes, it is possible to get an online quote as an instant cash offer for a trade-in; however, often these offers tend to play it safe in the valuations quoted to compensate for not knowing the actual condition of one's vehicle. Any of the tools that do go more in-depth to determine the specific state of a trade, completing these forms, is overwhelming due to the number of questions to be answered.

One thing people have in common is they all love to take pictures, and no one likes completing long forms. What if instead, one clicks a link that guides them to capture a few photos of their trade, and artificial intelligence and machine learning algorithms can fill in all the blanks so they can move forward and complete their next purchase entirely online? Discovery Loft successfully brought this capability to market in Canada in November 2019 with PAVE.

According to Stephen, PAVE pushes the boundaries of computer vision when combined with its proprietary real-time human-in-the-loop supervision to impact more than just consumers who have trade-ins online. This same need for granular transparency of the condition of a vehicle that is validated by a third-party affects a vast number of other scenarios as well, like wholesale transactions between retailers at auctions and direct from dealer-to-dealer, logistics, car-sharing, and rentals.

PAVE has been developed with an API first approach that provides clients with the ability to interface with one click, its user-friendly mobile browser-based capture UI and automated inspection capabilities in any application, website or marketplace with minimum development required. PAVE's API can also be integrated into other photo capture solutions currently in use to expand their capabilities to include evaluations. The platform is set to move the industry forward in 2020, with opening many new opportunities for automating workflows and creating and connecting unique ecosystems for the industry.

More Than Just Classifying Vehicle Damage

Identifying various classes of damage on vehicles like a cracked windshield, scrapes, scratches, and dents using image recognition is nothing new. A developer today can quickly build this skill into their application using many off the shelf visual classifiers like IBM Watson and Amazon SageMaker along with many open-source repositories available on GitHub. 2020 is such an exciting time for AI, deep learning and computer vision because virtually anyone can quickly prototype an idea that shows well in their presentation. It does impress because it has some real functionalities. Stephen says, "However, once you dig a little deeper into your use-case (hopefully not after the plan gets sold), you'll discover that it is just that, a prototype. These available models are very limiting for a lot of reasons."

For a day, if people followed a professional vehicle inspector at an auction as they do cosmetic inspections on vehicles, they would be shocked at the number variations of the types of damage and blemishes they have to detect. PAVE needed to be able to extract the elements and objects from digital images for over 10,000 unique types of vehicle damages to match their trained skills and years of human experience. The next most significant limitation will be to identify the exact part of the vehicle that is affected. To automate vehicle inspections that have real value to the industry, customers need to be able to locate the point of interest at the component level, not just "has damage on the front or side of the vehicle." PAVE identifies the complete anatomy of the over 22,000 body types of cars, trucks, and vans sold in North America over the past 15 years, down to knowing the marker light on the side mirrors.

Discovery Loft's View of the Current and Future State of Image and Facial Recognition

Stephen believes the ability to automate workflows using Image and Face Recognition Applications in 2020 have proven their incredible value, and what can already be achieved using current-day technologies. As this technology and its approaches evolve into the future, it will require significantly less computation and time to train and interpret the objects in the surroundings from images and videos more accurately than humans, which will open an incredible number of tasks that can be augmented. Image and Face Recognition is already surpassing human capabilities to see because it doesn't share the human's limitation of only having two eyes and one brain, who can only see and process one picture/scene at a time. It can simultaneously look at that same scene from any vantage points required, and it can also view it from a massive amount of different scenes at one time, giving it a broader 360-perspective to see patterns in what it sees that are beyond any human. Computer vision is entering a time of unprecedented growth; unlike anyone has seen before from any invention.