ABSTRACT
An automatic vehicle damage detection platform can enhance the customer claiming process and reduce the unnecessary cost of repair for an insurance company. Typically, the claim estimation process is manual which requires human experts to evaluate the damage cost. This is error-prone, time-consuming, and requires man-hour workers. In this chapter, a damaged vehicle part detection platform, called Intelligent Vehicle Accident Analysis (IVAA) which provides artificial intelligence as a service (AIaaS), is proposed. The system helps automatically assess vehicle parts’ damage and severity level. An insurance company can utilize our service to speed up the claiming process. IVAA is built on the docker image which allows the system to be scaled depending on the workload efficiently. Capsule neural network (CapsNet) is applied for damage recognition including two phrases: damage localization and damage classification. The accuracy of the damage localization is 93.28% and the accuracy of the damage classification is 98.47%, respectively.
