Detecting Insurance Fraud: A Study on Field Fires with Computer Vision and IoT

Abstract

The article suggests an automated system for overseeing the fraud detection process related to insurance claims for field fires in agriculture. This innovative solution combines computer vision, deep learning, and the Internet of Things (IoT) to leverage the strengths of each technology. As far as our knowledge extends, such an integration of these technologies has not been previously employed for analyzing insurance fraud in agriculture. The model actively monitors input from IoT devices equipped with infrared and temperature sensors. When these sensor values surpass predefined thresholds, the IoT device captures images of the field. These images are then processed by a fire detection model trained with various classifiers, allowing for performance comparisons. The reported results indicate an impressive accuracy of 97%, with potential for further improvement through a refined dataset specifically tailored for fraud detection.