Analyzing customer videos to identify fraud propensity

Problem Statement


As part of the insurance claim process, customers are to upload a video testimonial to report the reason for the loss of the device they have insured. The requirement was to analyze the video and use non-verbal cues to detect any fraud indicators.




Video analytics using Python-based modules.


Experts opinion was used to define behavior exhibited when a person tends to lie.


Image and video analytics was used to detect those patterns and arrive on the probability of fraud.




Automation of this task enabled cost savings for the insurer. Reduction in fraud and a more efficient claims process