Deep Learning for Sustainable Aquaculture: Opportunities and Challenges DOI Open Access
Alex Wu,

Ke-Lei Li,

Ziying Song

и другие.

Sustainability, Год журнала: 2025, Номер 17(11), С. 5084 - 5084

Опубликована: Июнь 1, 2025

With the rising global demand for aquatic products, aquaculture has become a cornerstone of food security and sustainability. This review comprehensively analyzes application deep learning in sustainable aquaculture, covering key areas such as fish detection counting, growth prediction health monitoring, intelligent feeding systems, water quality forecasting, behavioral stress analysis. The study discusses suitability architectures, including CNNs, RNNs, GANs, Transformers, MobileNet, under complex environments characterized by poor image severe occlusion. It highlights ongoing challenges related to data scarcity, real-time performance, model generalization, cross-domain adaptability. Looking forward, paper outlines future research directions multimodal fusion, edge computing, lightweight design, synthetic generation, digital twin-based virtual farming platforms. Deep is poised drive toward greater intelligence, efficiency,

Язык: Английский

Explainable models for predicting crab weight based on genetic programming DOI Creative Commons
Tao Shi,

Lingcheng Meng,

Limiao Deng

и другие.

Ecological Informatics, Год журнала: 2025, Номер unknown, С. 103131 - 103131

Опубликована: Апрель 1, 2025

Язык: Английский

Процитировано

0

Underwater Weight Estimation of Three Sea Cucumber Species in Culture Tanks Using Image Analysis and ArUco Markers DOI Creative Commons
Roongparit Jongjaraunsuk,

Saroj Rermdumri,

Kanokwan Khaodon

и другие.

Animals, Год журнала: 2025, Номер 15(8), С. 1121 - 1121

Опубликована: Апрель 13, 2025

Sea cucumbers play a vital role in marine and coastal ecosystems, with some species holding significant economic value. Accurate growth assessment, particularly weight estimation, is crucial for their management conservation. However, direct measurement poses challenges, as sea expel internal fluids when handled, altering body size weight. This study evaluates the effectiveness of image analysis combined ArUco markers to estimate three economically ecologically important cucumber found Thailand: black (Holothuria leucospilota), pink warty (Cercodemas anceps), sandfish scabra). The proposed method demonstrated high accuracy, R2 values 0.9699, 0.9774, 0.9882, respectively. Furthermore, no differences (p > 0.05) were observed between traditional hand measurements image-based assessments, relative errors 7.71 ± 4.30% cucumber, 5.06 3.37% 4.50 3.23% sandfish. Unlike deep learning, which requires large datasets computation, this simple, cost-effective, adaptable highlights potential non-invasive accurate tool estimating approach minimizes stress on animals can be extended other aquatic species. challenges such shadows, water turbidity, presence similarly shaped objects near should considered applying technique field conditions.

Язык: Английский

Процитировано

0

100 Years of Penaeid Domestication and Meta-Analysis of Breeding Traits DOI
Shengjie Ren, José M. Yáñez, Ricardo Pérez-Enríquez

и другие.

Reviews in Fisheries Science & Aquaculture, Год журнала: 2025, Номер unknown, С. 1 - 20

Опубликована: Май 7, 2025

Язык: Английский

Процитировано

0

Deep Learning for Sustainable Aquaculture: Opportunities and Challenges DOI Open Access
Alex Wu,

Ke-Lei Li,

Ziying Song

и другие.

Sustainability, Год журнала: 2025, Номер 17(11), С. 5084 - 5084

Опубликована: Июнь 1, 2025

With the rising global demand for aquatic products, aquaculture has become a cornerstone of food security and sustainability. This review comprehensively analyzes application deep learning in sustainable aquaculture, covering key areas such as fish detection counting, growth prediction health monitoring, intelligent feeding systems, water quality forecasting, behavioral stress analysis. The study discusses suitability architectures, including CNNs, RNNs, GANs, Transformers, MobileNet, under complex environments characterized by poor image severe occlusion. It highlights ongoing challenges related to data scarcity, real-time performance, model generalization, cross-domain adaptability. Looking forward, paper outlines future research directions multimodal fusion, edge computing, lightweight design, synthetic generation, digital twin-based virtual farming platforms. Deep is poised drive toward greater intelligence, efficiency,

Язык: Английский

Процитировано

0