Lobster Growth Monitoring with AI-Based Computer Vision Using SVM and Neural Network

Abstract

A modern agricultural and aquaculture technique today heavily relies on computer assistance. Computers aid in the analysis, identification, and regulation of feeding patterns, making the process more effective. For example, lobster farming is now predominantly conducted using pond-based methods, known as aquaculture, rather than sourcing lobsters from the wild. This is because lobsters are highly sensitive creatures, and failing to replicate their natural habitat can lead to crop failure. Several factors influence lobster farming conditions, including water quality, feed quantity, and lobster species. Another critical factor is disease outbreaks, which can spread rapidly due to the high lobster density in a single pond. Managing these conditions manually is impractical due to the large number of ponds and the need to replicate natural habitat conditions accurately. To address these challenges, a monitoring mechanism utilizing artificial intelligence (AI)-based image processing is implemented. AI methods can manipulate environmental conditions to closely resemble a lobster’s natural habitat by monitoring pH levels, determining gender, and assessing health status. Data accuracy is ensured using two algorithmic approaches. Experimental results show that the application is designed as a GUI with simple features, making it user-friendly for farmers and the general public. This application was tested using a sample of 200 lobsters, achieving a data accuracy rate of 95% with the SVM algorithm and 85% with the Neural Network algorithm. The application can identify lobster species, size, and potential diseases affecting them.