Utilising computer vision techniques for automated density and growth estimation in precision aquaculture systems for prawn cultivation

Lean, Jin Hao (2025) Utilising computer vision techniques for automated density and growth estimation in precision aquaculture systems for prawn cultivation. Final Year Project, UTAR.

PDF Download (3401Kb)

Abstract

Prawn farming, a vital sector of the global aquaculture industry, faces challenges with traditional monitoring methods that are labor-intensive, error-prone, and lack real-time capabilities, leading to inefficiencies in feeding and harvest planning, particularly for small- and medium-scale farmers. This project aims to address these issues by developing a computer vision-based system for automated density and growth estimation of Cherax quadricarinatus prawns, enhancing operational efficiency and sustainability. Utilizing the lightweight YOLO11n neural network, a Raspberry Pi 5, and a PiCamera (Night Vision), the system automates prawn monitoring, improves accuracy through machine learning, and ensures affordability at $60-$80 per unit. A Cron Job feature enables continuous data collection, building a farm-specific dataset to overcome the lack of standardized prawn data. Deployed in a controlled pond environment, the system captured 2000 images under varying conditions, achieving real-time detection at 5 FPS, though initial tests revealed accuracy issues requiring further data and fine-tuning. By mitigating challenges like environmental variability, high costs, and technical complexity identified in prior studies, this solution offers a scalable, user-friendly tool that empowers smaller farms to optimize resource use and enhance productivity in precision aquaculture.

Actions (login required)