Machine Vision Usage in Insurance

Insurers now have access to an unprecedented quantity of image and video data and are beginning to invest in machine vision to process it.

Insurers now have access to an unprecedented quantity of image and video data. Many still manually review these data sources, but this provides limited insight. Carriers are beginning to invest in machine vision technology to process this data, programmatically analyzing risk factors and making sense of these vast image stores. Machine Vision: What Is It? Machine vision is the AI-based analysis of images from sources like smartphone photos, drones, low-lying aircraft, satellites and dashcams. Machine vision platforms offer analysis—i.e., the ability to upload images from a proprietary source into a platform—or they can be trained from scratch to work with an insurer’s business. Dedicated platforms can provide a relatively lightweight way to help insurers automate, scale and enhance risk evaluation while seeing gains in operational efficiency and cost reduction. The Move to Purpose-Built Platforms General machine learning platforms may be capable of image- and video-based analysis of risk factors in the not-too-distant future. Yet, for the time being, insurers are likely to see more tangible results by implementing a machine vision platform built specifically for insurance needs in claims and underwriting. These solutions are likely to provide more value with fewer resources and less investment. Some purpose-built machine vision solutions for the insurance industry may use general-purpose platforms from other providers behind the scenes. But the insurance-focused vendors have done the work of training solutions for specific insurance use cases so that insurers don’t have to. See also: Rise of the Machines in Insurance   Machine Vision Use Cases Most current machine vision use cases focus on commercial and personal property underwriting and claims due to the proliferation of property imagery, especially for roof analysis. Usage is emerging for auto claims, where the predominant application is claims damage and estimation. Machine vision is mostly exploratory in other lines of business; one emerging example is life insurance, in which machine vision can perform image analysis to aid in underwriting. Use of images to determine claims and underwriting risk factors isn’t necessarily a new concept for insurers; underwriters have been using sources like Google satellite images for years for this precise purpose. Yet unstructured sources of photo and video data continue to proliferate, and machine vision can help insurers evaluate a broader range of risk and automate decision-making. More information on the space is available in Novarica’s latest report, Machine Vision in Insurance: Use Cases and Emerging Providers, which provides an overview of machine vision technology as well as prominent vendors.

Jeff Goldberg

Profile picture for user JeffGoldberg

Jeff Goldberg

Jeff Goldberg is head of insurance insights and advisory at Aite-Novarica Group.

His expertise includes data analytics and big data, digital strategy, policy administration, reinsurance management, insurtech and innovation, SaaS and cloud computing, data governance and software engineering best practices such as agile and continuous delivery.

Prior to Aite-Novarica, Goldberg served as a senior analyst within Celent’s insurance practice, was the vice president of internet technology for Marsh Inc., was director of beb technology for Harleysville Insurance, worked for many years as a software consultant with many leading property and casualty, life and health insurers in a variety of technology areas and worked at Microsoft, contributing to research on XML standards and defining the .Net framework. Most recently, Goldberg founded and sold a SaaS data analysis company in the health and wellness space.

Goldberg has a BSE in computer science from Princeton University and an MFA from the New School in New York.


Read More