Litcius/Paper detail

Car Damage Detection and Assessment Using CNN

Atharva Shirode, Tejas Rathod, Parth Wanjari, Aparna Halbe

20222022 IEEE Delhi Section Conference (DELCON)20 citationsDOI

Abstract

In today's digital world, most businesses are adopting technology in every possible way. Many time it occurs that when the car is damaged insurance claims are done. If the car is insured, a person from the insurance industry visits and takes survey of the customers car and prepares the report. The manual verification is a tedious process. But with the major advancement in field of deep learning algorithms, it can be used in the insurance industry to solve these problems. In the proposed solution we have implemented 2 CNN models. VGG16 is used to detect the damage on the car, location of the damage and its severity. Mask RCNN is used to mask out the exact damaged region. Both the models give a fair idea about the damage caused to the car which can help insurance company to proceed further with the insurance claims without wasting time and resources on manual verification.

Topics & Concepts

Computer scienceProcess (computing)Automobile insuranceField (mathematics)Insurance industryDeep learningArtificial intelligenceRisk analysis (engineering)Computer securityBusinessActuarial scienceOperating systemPure mathematicsMathematicsAdvanced Neural Network ApplicationsVehicle License Plate RecognitionAutonomous Vehicle Technology and Safety
Car Damage Detection and Assessment Using CNN | Litcius