Towards classification of shrimp diseases using transferred convolutional neural networks

Abstract

Vietnam is one of the top 5 largest shrimp exporters globally, and Mekong delta of Vietnam contributes more than 80% of total national production. Along with intensive farming and growing shrimp farming area, diseases are a severe threat to productivity and sustainable development. Timely response to emerging shrimp diseases is critical. Early detection and treatment practices could help mitigate disease outbreaks, leading to on-site diagnostics, instant services recommendation, and front-line treatments. The authors establish a contribution hub for data collection in the ethnographic fieldwork of Mekong delta. Several deep convolutional neural networks are trained by applying the transfer learning technique. We have investigated six common reported shrimp diseases. The classification accuracy is achieved of 90.02%, which is very useful in extremely non-standard images. Throughout the work, we raise the attention of shrimp experts, computer scientists, treatment agencies, and policymakers to develop preventive strategies against shrimp diseases.