SmartInsure: Blockchain and CNN Leveraged Secure and Efficient Cattle Insurance

Agriculture is one of the significant industries that encompasses various activities including crop cultivation and livestock farming to provide necessary resources to societies. Along with providing necessary resources for global stability and sustenance it also serves as the backbone for several countries economies. Agriculture is highly dependent on unpredictable weather and climatic events, making mitigation techniques futile. Due to these uncertainties, many farmers face various hardships when attempting to meet the expected yield and finances. Another major factor influencing livestock farming, especially livestock, is the spread of viral diseases like Bovine Respiratory Disease (BRD) and Foot-and-Mouth Disease (FMD). Livestock insurance is a specific type of insurance within broader agriculture insurance that serves several critical purposes for farmers such as risk mitigation, financial stability, Investment confidence, etc. Currently employed cattle insurance systems are centralized and involve multiple parties exchanging information and payments. These entities include farmers, insurers, veterinarians, and regulatory agencies which are geographically apart and cause significant delays and administrative costs. Hence, a Blockchain and Convolution Neural Network (CNN) leveraged solution SmartInsure is proposed to increase transparency, data integrity, and efficiency, along with a fool-proof way of identifying the insured cattle. The identification of cattle is a major problem in cattle insurance and the lack of a robust identification system can lead to insurance fraud. Hence CNN-based cattle identification using muzzle images can be an effective solution for avoiding such conflicts while processing insurance claims. Considering the cost-effectiveness of the proposed SmartInsure, off-chain distributed data storage is employed for storing muzzle images of insured cattle. Proposed SmartInsure Proof-of-Concept (POC) is designed, and functional analysis is performed to check the business logic implemented. Along with the functional analysis, the efficiency of the CNN model is evaluated for validation accuracy and loss metrics. Results from the analysis show that the cattle are identified with 94.11% validation accuracy and 0.38 validation loss.