Automatic Counting of Shrimp Larvae using Artificial Intelligence

Abstract

The research is to revolutionize the shrimp farming industry by developing a computer vision and Artificial Intelligence (AI) system for accurate and efficient shrimp counting. Recently, the aquaculture industry plays important roles in the global demand for seafood. In particular, shrimp farming has become a significant contributor to the industry. However, the process of counting the shrimp larvae is a labour-intensive and time-consuming task that often requires manual effort, leading to inefficiencies, inaccuracies and increased operational costs. Thus, the project was conducted to challenge the shrimp larvae counting with robust and efficient method. The proposed system captures and analyses images of Macrobrachium Rosenbergii shrimp post larvae (PLs) of varying stages and quantities using a high-resolution webcam and a Convolutional Neural Network framework (CNN). The presence of molts, feed and debris in the imaging chamber is considered by the system. The goal is to have a low mean absolute error when counting large and small PLs. This technology's successful implementation will not only improve the accuracy and reliability of shrimp counting but will also clear the way for counting other small aquatic species in their larval stages, such as fish, crabs, oysters and eggs. The methodology of the project entails training the system with a large image dataset and testing its performance with the trained model. The development of a fast and precise shrimp counting AI system, which has the potential to revolutionize the industry and improve customer satisfaction, is one of the significant results and findings. Finally, this study proposes a different approach to automate counting shrimp larvae counting by combining computer vision and AI, providing a more accurate, efficient and reliable solution for the aquaculture industry.