A Survey of Deep Learning for Intelligent Feeding in Smart Fish Farming DOI
Peng Xiao-hong, Tianyu Zhou,

Zhenlu Wu

и другие.

Опубликована: Ноя. 24, 2023

The rapid development of deep learning has been successfully applied in various fields, including aquaculture, providing new methods and ideas to realize unmanned intelligent aquaculture. This paper focuses on the technology research used for feeding fish farming past decade, discusses application detail, behavior analysis, detection tracking live fish, growth state monitoring, residual bait identification counting, water quality prediction, etc., summarizes evaluates methods, at same time, analyzes technical details precision is analyzed details, data, algorithms, evaluation performance indexes. summarized results show that advantage lies automatic extraction features, which also provides support construction system. However, due large differences species, aquaculture environments data acquisition less portable, it still stage weak artificial intelligence, requires a amount train model, cost high, become bottleneck restricts further Nevertheless, made breakthroughs processing complex data. In summary, purpose this review provide researchers producers with better understanding current status strong theoretical production process.

Язык: Английский

Piscicultura inteligente: a integração das Tecnologia 4.0 e “Business Intelligence” para gestão ágil na aquicultura DOI Creative Commons
Thales Francisco Gonçalves, Johana Marcela Concha Obando, Luiz Cláudio Chiavani Júnior

и другие.

Revista de Gestão e Secretariado (Management and Administrative Professional Review), Год журнала: 2025, Номер 16(1), С. e4524 - e4524

Опубликована: Янв. 9, 2025

A crescente demanda por alimentos devido ao aumento populacional pressiona a pesca de captura e esgota os estoques peixes. Como alternativa, aquicultura avançada surge, embora ainda não tenha alcançado o mesmo nível tecnológico outros setores. está em crescimento, espera-se que até 2030 forneça maior parte do peixe consumido globalmente. No entanto, setor, muitas partes mundo, enfrenta desafios. As tecnologias 4.0, podem proporcionar ferramentas para criação pisciculturas inteligentes, usam Internet of Things, big data, Inteligência Artificial blockchain promover eficiência sustentabilidade. Neste contexto, Business Intelligence (BI) aparece como uma alternativa essencial auxiliar transformação data conhecimento gestores tomadores decisão na aquicultura. Esta revisão tem objetivo explorar conceitos (BI), piscicultura inteligente digitais aplicadas à aquicultura, proporcionando visão atualizada dos avanços área. Para atingir esse objetivo, foram analisadas cinco revisões recentes sobre estado atual das 4.0. Além disso, busca sistemática resultou coleta 20 artigos originais adicionais. O presente trabalho oferece organizada estudos abordam inteligente, tempo integra algumas pesquisas focadas aplicação conceito BI. Os trabalhos analisados destacam informações chave ser integradas piscicultura, no Brasil globalmente, com facilitar tomada decisões gestão sustentável

Процитировано

0

An efficient detection model based on improved YOLOv5s for abnormal surface features of fish DOI Creative Commons
Zheng Zhang, Xiang Lu, Shouqi Cao

и другие.

Mathematical Biosciences & Engineering, Год журнала: 2024, Номер 21(2), С. 1765 - 1790

Опубликована: Янв. 1, 2024

Detecting abnormal surface features is an important method for identifying fish. However, existing methods face challenges in excessive subjectivity, limited accuracy, and poor real-time performance. To solve these challenges, a accurate detection model of in-water fish proposed, based on improved YOLOv5s. The specific enhancements include: 1) We optimize the complete intersection over union non-maximum suppression through normalized Gaussian Wasserstein distance metric to improve model's ability detect tiny targets. 2) design DenseOne module enhance reusability features, introduce MobileViTv2 speed, which are integrated into feature extraction network. 3) According ACmix principle, we fuse omni-dimensional dynamic convolution convolutional block attention challenge extracting deep within complex backgrounds. carried out comparative experiments 160 validation sets fish, achieving precision, recall, mAP

Язык: Английский

Процитировано

3

Overview of aquaculture Artificial Intelligence (AAI) applications: enhance sustainability and productivity, reduce labor costs, and increase the quality of aquatic products DOI Open Access

Sherine Ragab,

Seyed Hossein Hoseinifar, Hien Van Doan

и другие.

Annals of Animal Science, Год журнала: 2024, Номер unknown

Опубликована: Авг. 29, 2024

Abstract The current work investigates the prospective applications of Artificial Intelligence (AI) in aquaculture industry. AI depends on collecting, validating, and analyzing data from several aspects using sensor readings, feeding sheets. is an essential tool that can monitor fish behavior increase resilience quality seafood products. Furthermore, algorithms early detect potential pathogen infections disease outbreaks, allowing stakeholders to take timely preventive measures subsequently make proper decision appropriate time. predict ecological conditions should help farmers adopt strategies plans avoid negative impacts farms create easy safe environment for production. In addition, aids analyze collect regarding nutritional requirements, nutrient availability, price could adjust modify their diets optimize feed formulations. Thus, reduce labor costs, aquatic animal’s growth, health, formulation waste output detection outbreaks. Overall, this review highlights importance achieve sustainability boost net profits

Язык: Английский

Процитировано

2

State of charge estimation for lithium‐ion battery pack based on real vehicle data and optimized backpropagation method by adaptive cross mutation sparrow search algorithm DOI Creative Commons
Juqiang Feng,

Feng Cai,

Long Wu

и другие.

Energy Science & Engineering, Год журнала: 2023, Номер 12(3), С. 896 - 912

Опубликована: Дек. 25, 2023

Abstract In response to the issues of traditional backpropagation (BP) neural networks in state charge (SOC) estimation, including easy convergence local optima, slow speed, and low accuracy, this paper proposes a novel adaptive crossover mutation strategy dynamic sparrow search algorithm optimize BP networks' initial values thresholds (ACMSSA‐BP). The proposed method is based on algorithm, where number producers scroungers adjusted through an factor. This improvement effectively transitions process from extensive full exploration localized fine‐tuning search. position update phase producers, strategies are introduced increase diversity good populations, prevent converging maintain its capability later stage. Using real transportation data coal mining flame‐proof tracked vehicles, we applied correlation theory extract model feature parameters constructed training set estimate SOC. results both static validation experiments have indicated that ACMSSA‐BP has delivered impressive performance predicting SOC, as reflected mean absolute error, root squared percentage error less than 1.5%, 1.6%, respectively. Compared with BP, SSA‐BP, CMSSA‐BP, PSO‐BP, NARX_NN methods, approach demonstrates enhanced accuracy significant robustness, generalization capabilities.

Язык: Английский

Процитировано

5

A Survey of Deep Learning for Intelligent Feeding in Smart Fish Farming DOI
Peng Xiao-hong, Tianyu Zhou,

Zhenlu Wu

и другие.

Опубликована: Ноя. 24, 2023

The rapid development of deep learning has been successfully applied in various fields, including aquaculture, providing new methods and ideas to realize unmanned intelligent aquaculture. This paper focuses on the technology research used for feeding fish farming past decade, discusses application detail, behavior analysis, detection tracking live fish, growth state monitoring, residual bait identification counting, water quality prediction, etc., summarizes evaluates methods, at same time, analyzes technical details precision is analyzed details, data, algorithms, evaluation performance indexes. summarized results show that advantage lies automatic extraction features, which also provides support construction system. However, due large differences species, aquaculture environments data acquisition less portable, it still stage weak artificial intelligence, requires a amount train model, cost high, become bottleneck restricts further Nevertheless, made breakthroughs processing complex data. In summary, purpose this review provide researchers producers with better understanding current status strong theoretical production process.

Язык: Английский

Процитировано

0