Behavioral Analysis For Shrimp Feeding Optimazation

Dave, Michael (2025) Behavioral Analysis For Shrimp Feeding Optimazation. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5022211174-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB) | Request a copy

Abstract

Produksi udang di Malaysia terus meningkat, dengan pakan menyumbang sekitar 60% dari total biaya produksi. Namun, pemberian pakan berlebih telah menjadi masalah utama karena pakan yang tidak termakan dapat menurunkan kualitas air. Dalam budidaya udang, kondisi ini dapat menyebabkan kematian. Oleh karena itu, pemantauan yang efisien menjadi sangat penting. Metode tradisional masih mengandalkan observasi manual, yang bersifat memakan waktu dan subjektif. Penelitian terkini menunjukkan bahwa metode deep learning, seperti model YOLO, mampu mendeteksi berbagai objek untuk analisis lanjutan, termasuk analisis perilaku. Proyek ini bertujuan untuk mengembangkan sistem deteksi udang dan analisis perilaku guna memprediksi tingkat kelaparan udang. Sistem deteksi mencapai nilai recall sebesar 94,17%, yang menunjukkan bahwa model mampu mendeteksi udang secara benar dengan jumlah false negative yang rendah. Analisis perilaku berhasil mengklasifikasikan tingkat kelaparan dengan akurasi antara 80% hingga 89,2%. Namun, keterbatasan metode ini adalah kesulitan model dalam membedakan tingkat kelaparan menengah akibat adanya perilaku yang saling tumpang tindih . Hasil ini menunjukkan potensi besar dari kombinasi antara deteksi objek dan analisis perilaku dalam sistem pemantauan budidaya udang.
======================================================================================================================================
Shrimp production in Malaysia is increasing, with feed contributing to 60% of the total production cost. However, overfeeding has become a major concern as uneaten feed can affect the quality of the water. In a shrimp case, this factor could lead to death. Therefore, efficient monitoring is important. The traditional method relies on manual observation, which is timeconsuming and subjective. Recent research has shown that deep learning methods, such as the YOLO model, can detect various objects for extended analysis, such as behaviour. This project aims to develop a shrimp detection and behavioural analysis to predict the shrimp’s hunger level. The detection achieved a recall of 94.17%, which means that the model managed to detect a true shrimp with few false negatives. The behaviour analysis successfully classified hunger level, achieving accuracy between 80% and 89.2%. However, the limitation of this method is that the model has difficulty differentiating the intermediate level due to mixing behaviour. This result shows the potential of the combination between object detection and behavior analysis.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Aquaculture, UDang, Perilaku, Computer Vision, YOLO, Optical Flow, Kelaparan, Pemberian makan terhadap udang, Aquaculture, Shrimp, Behaviour, Computer vision, YOLO, Optical flow,Hunger, Feeding shrimp Aquaculture, Shrimp, Behaviour, Computer vision, YOLO, Optical flow, Hunger, Feeding shrimp
Subjects:	S Agriculture > SF Animal culture S Agriculture > SH Aquaculture. Fisheries. Angling T Technology > T Technology (General)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Michael Christian Dave
Date Deposited:	28 Jan 2026 07:41
Last Modified:	28 Jan 2026 07:41
URI:	http://repository.its.ac.id/id/eprint/130810

Actions (login required)

View Item