A leading insurance company had an expensive and laborious process to inspect rooftop damages
The company was looking for a comprehensive drone imagery analytics solution that could automate damage assessment and claims estimation reporting while minimizing risks
WNS deployed an advanced image analytics solution underpinned by artificial intelligence and machine learning to improve operational efficiency
Insurance companies have to deploy many resources to process and verify claims related to rooftop damages, ranging from physical surveys to manual checking of the photographs. In the case of roofs on commercial and residential buildings, the area to be surveyed could be up to 50,000 square feet in size. The process then becomes time-consuming and laborious. Until recently, manual inspections — which are inherently marred by inaccuracies and safety concerns — were the only options available to verify such claims.
Video and image analytics have resulted in enhanced drone imagery solutions to generate significant value for insurers. From risk assessments and automated claim settlements to detecting false claims, this technology is already driving measurable business impact.
Deploying costly resources to physically verify the rooftop damages caused by weather-related incidents such as hailstorms. Other challenges for the company included limited roof access, safety and liability concerns, and inaccuracies in manual inspections. Moreover, the available drone imagery solutions were either unable to provide correct labeling of damaged / non-damaged rooftops, or extract relevant data from high-resolution images. Another major concern was getting the appropriate infrastructure with high-end parallel processing computers.
The company was looking for a comprehensive drone imagery analytics solution that could automate damage assessment and claims estimation reporting while minimizing risks.
The insurer required a robust infrastructure platform to handle, process and extract actionable insights on claims assessment to improve operational efficiency, expedite customer service and increase long-term return on investment.
WNS deployed an automated Artificial Intelligence (AI)- and Machine Learning (ML)-based drone imagery analytics solution. Backed by a robust deep-learning methodology, the solution was built to solve the most complex challenges in image processing, cleansing, segmentation and feature extraction.
The solution included three major components:
Identifying rooftop damages via a combination of image pre-processing, brightness normalization, image thresholding and contour / edge detection for better clarity; image segmentation and feature extraction using supervised ML and deep-learning algorithms
Analyzing past information on damaged rooftops by surveyors and historical claims assessment value
Triangulating image, claims and historical data using ensemble modeling techniques to automate the identification of risks and damage assessments
By combining structured and unstructured claims data with drone-captured images and other data sources, the algorithm automatically locates and classifies damages due to hailstorm and other weather incidents. It then provides various metrics to help in appropriate claims decisions.
That a robust, advanced analytics-powered image classification methodology can re-shape claims assessment and minimize risk factors, leading to increased operational efficiency and improved customer service. Other benefits include:
95 percent accuracy in damage prediction, leading to elimination of manual effort and time spent segregating images
Easing the process of assessing the validity of claims, identifying risk factors and automating claims estimation reporting
The WNS drone imagery solution offers potential benefits of more than USD 30 Million through automation and increased operational efficiency.
Our proven model development, automated validation process and robust evaluation metrics continue to enable the client to prioritize inspections, proactively measure damages and reduce the claims cycle time.
Join the conversation
30 March 2023
Finance and Accounting
23 February 2022
25 January 2022