Toward Aquaculture Automation: Image Enhancement and Containment Net Removal for Robust Fish Detection for Behavior Recognition
2025 IEEE Ocean Engineering Technology and Innovation Conference, OETIC 2025, Hybid, Purwokerto, Endonezya, 16 Aralık 2025, ss.157-162, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/oetic67587.2025.11512868
- Basıldığı Şehir: Hybid, Purwokerto
- Basıldığı Ülke: Endonezya
- Sayfa Sayıları: ss.157-162
- Anahtar Kelimeler: Aquaculture, fish detection, image enhancement, Sobel kernel
- Orta Doğu Teknik Üniversitesi Kuzey Kıbrıs Kampüsü Adresli: Evet
Özet
Aquaculture plays a vital role in global food security by providing controlled environments for raising aquatic animals. Effective and precise feeding management is essential for sustainability; however, traditional practices rely heavily on farmers’ judgment. This judgment varies among veterans and novices, leading to food wastage, poor fish health, and inadequate monitoring of fish health. To address these issues, automation offers a promising solution for accurate control of feeding operations. A vision system is essential. However, a significant challenge in underwater environments arises owing to variable lighting, light backscattering, and physical obstructions such as containment nets. These factors hinder the accurate observation of fish behavior, which is crucial for feeding strategy development. This study presents an image-processing algorithm designed to eliminate visual obstructions caused by the net structures in underwater environments. Our approach achieved a significantly high precision for fish (98.8%). This enabled a better understanding of fish behavior and led to more reliable automated feeding systems. These findings highlight the potential of computer vision techniques in aquaculture monitoring, promising to improve operational efficiency and reduce reliance on manual labor.