Simplifying claims processing by integrating digital technologies to accelerate operations
Client Overview
The client is an international insurtech with a mission to build the world’s leading protection and technology-enabled insurance ecosystem. Our client serves customers in 26 markets across North America, Asia and Europe.
The Problem
The client did not have an automated service to identify screen damage in the mobile phone image uploaded by the claimant to check the validity of their claim. This led to an increase in the claims cycle significantly which often led to delays in the settlement process. The end customers had to often visit customer care centers and insurance offices to buy or claim the mobile insurance, which was time consuming and hectic. There was a need for a one-stop solution that would be easily accessible to customers and would help them get insurance in a hassle-free manner.
Our Solution
The company engaged with PureSoftware to develop an automated service to identify screen damage in the mobile phone image uploaded by the claimant. PureSoftware accelerated the claims process by achieving 85% accuracy in identifying the screen damage.
Our First Step, PureSoftware team of experts developed and implemented a Python based custom algorithm for Image Processing using sci-kit & OpenCV2 image library. They also implemented CNN based VGC16 (TensorFlow & Keras) ML algorithm in Python. Next, the team developed an object detection technology for auto capturing of mobile screen images using Tracker.JS technology. Integration with Flask based Restful APIs was done using python language & hosted on Azure VM (IaaS). The developed solution by PureSoftware was made available on mobile native & hybrid apps.
The Result
- Saved end customers’ time to visit customer care centres or insurance office to buy or claim the mobile insurance.
- 80-85% accuracy achieved through the machine learning model.
- Automated service to detect damages accelerated the claims process significantly.
- 80-85% accuracy achieved through the machine learning model.
Technology Stack
- Azure SQL
- Anaconda Python 3.4
- Sci-Kit Image library
- NumPy
- OpenCV2
- TensorFlow & Keras
- Machine Learning Model VGC16 (CNN)
- Tracker.JS
- Angular 4
- Flask APIs
