Integrating UAVs and AI for shrimp pond mapping: A case study in Can Giuoc district, Long An province, Vietnam

Abstract

Worldwide, the utilization of Unmanned Aerial Vehicles (UAVs) has been de-ployed in a wide range of resources management practices. The UAVs serve as valuable and visible tools for managing water resources, forests, agriculture, and land use change. Undoubt-edly, the application of UAVs surpasses traditional methods in terms of efficiency, offering significant time and cost savings. Meanwhile, Artificial intelligence (AI) has emerged as a critical technology in the realm of information technology, particularly when it comes to image segmentation. The purpose of this study is to integrate UAVs and AI for mapping shrimp farms in Long An province, Mekong Delta. By leveraging AI, we empower systems to learn intricate image features and subsequently identify and segment objects within those images. In the con-text of modern agricultural management practices, we leverage UAV imagery as input data for AI systems to identify shrimp ponds, the image recognition platform is Deep Learning (DL) based on U – Net structure. Using the shrimp pond boundary on the 1:1000 scale topographic map as reference data, the results of this method showed a recall of 83.3%, corresponding to a miss rate of 16.7%. The precision of the method was 85.7%, corresponding to a misidentified shrimp pond extraction rate of 14.3%. The results suggest that the combination of UAVs and AI to mapping shrimp farms can facilitate efficient monitoring and management practices for local authorities. Thus, this integration is a promising application to assist and enable agri-cultural planning and regional economic development activities.