YOLO V8 FOR ESTIMATION OF SHRIMP BODY WEIGHT FROM IMAGES

Abstract

Penaeus vannamei, a highly cultured species, accounted for 51.7% of the total shrimp production, reaching 5.8 million tonnes globally in 2020 (FAO, 2022). Despite its substantial production, the shrimp farming industry faces various challenges, including shrimp growth monitoring, which is a critical aspect of production. Monitoring shrimp growth not only determines the rate of growth but also affects feeding efficiency. Currently, shrimp growth rate assessment relies on traditional methods that calculate average values from sampled data, introducing potential biases and necessitating time-consuming processes, such as the drying of shrimps before scaling takes place. To address these limitations, this research proposes a novel approach integrating image processing and machine learning algorithms to estimate shrimp weight. Specifically, we combine the YOLO V8 detection algorithm with logarithmic regression. YOLO V8 detects shrimp and measures their height in images, and then, utilizing the detected shrimp objects, we predict their weight through logarithmic regression. Our proposed algorithm achieves a Mean Average Error (MAE) of 0.6 grams in shrimp weight estimation, providing a more efficient and accurate alternative for shrimp growth monitoring in the industry.