Scholarlink

<p>Classifying UWI (Under Water Images) is a challenging process due to the complex underwater nature and the illumination conditions. Marine preservationists and scientists find it crucial to regularly assess the species in their ecosystems and track population variations. While labor-intensive manual analysis has traditionally been employed, several CAD (Computer Aided Designs) for automated WI have been introduced. Nevertheless, the absence of an optimum model for automated WI classification persists. This challenge arises primarily from the complexities inherent in UWI, including variations in environmental illumination, less resolution, dynamic backgrounds and water turbidity among certain organisms. To address these, challenges, this work presents a SAE (Stacked Auto Encoder) based A-GEO (Adaptive Golden Eagle Optimizer) for UWI classification. The SAE is used for learning high level features of UWI and removing redundant features. Then, the A-GEO is used for optimizing the weights and bias of the SAE, ensuring that the SAE learns the efficient representations of the UWI. This work is evaluated on the two UWI datasets and achieved better accuracies of 98.7% (Dataset 1) and 98.6% (Dataset 2). The conducted experiments demonstrate that the proposed system exhibits precise classification of large scale UWI with a notable level of accuracy, surpassing conventional models.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhanced underwater image classification using SAE-based adaptive golden eagle optimizer for high-accuracy marine species identification

  • 期刊论文

  • 发表日期：

    2025年2月5日

查看全文
  • Absa S,
  • Saju P John,
  • Y Mary Reeja,
  • Jarin T

摘要

Classifying UWI (Under Water Images) is a challenging process due to the complex underwater nature and the illumination conditions. Marine preservationists and scientists find it crucial to regularly assess the species in their ecosystems and track population variations. While labor-intensive manual analysis has traditionally been employed, several CAD (Computer Aided Designs) for automated WI have been introduced. Nevertheless, the absence of an optimum model for automated WI classification persists. This challenge arises primarily from the complexities inherent in UWI, including variations in environmental illumination, less resolution, dynamic backgrounds and water turbidity among certain organisms. To address these, challenges, this work presents a SAE (Stacked Auto Encoder) based A-GEO (Adaptive Golden Eagle Optimizer) for UWI classification. The SAE is used for learning high level features of UWI and removing redundant features. Then, the A-GEO is used for optimizing the weights and bias of the SAE, ensuring that the SAE learns the efficient representations of the UWI. This work is evaluated on the two UWI datasets and achieved better accuracies of 98.7% (Dataset 1) and 98.6% (Dataset 2). The conducted experiments demonstrate that the proposed system exhibits precise classification of large scale UWI with a notable level of accuracy, surpassing conventional models.