Automated density and growth estimation in precision aquaculture systems for prawn cultivation using computer vision techniques

Chong, Xiao Wei (2024) Automated density and growth estimation in precision aquaculture systems for prawn cultivation using computer vision techniques. Final Year Project, UTAR.

Preview

PDF Download (4Mb) | Preview

Abstract

Prawn cultivation is a crucial aquaculture industry, but it faces significant challenges related to inefficient feeding practices and lack of accurate population monitoring. Overfeeding due to imprecise population estimates leads to wasted resources and potential environmental issues. Additionally, traditional methods of monitoring prawn growth and well-being in underwater environments are labor-intensive and prone to inconsistencies, hindering timely decision-making processes. To address these challenges, this project proposes an innovative solution that leverages computer vision and machine learning techniques. By employing the state-of-the-art You Only Look Once (YOLO) v7 object detection algorithm, the project aims to develop a system capable of accurately detecting and classifying prawns based on their growth stages. The detected prawns are then measured, and their lengths are used to estimate their weights and categorize them into juvenile, subadult, or adult stages. Furthermore, the project automates the estimation of prawn density and population within the aquaculture system, providing farmers with valuable insights into the population dynamics. This automated approach eliminates the need for manual monitoring and enables more efficient resource allocation and management strategies. By addressing the challenges, this study contributes to the advancement of precision aquaculture operations. The proposed solution offers a viable path towards sustainable and efficient prawn farming practices, optimizing resource utilization, minimizing environmental impact, and ultimately enhancing the profitability and sustainability of the prawn cultivation industry.

Actions (login required)