Scholarlink

<p>Seaweed beds play a crucial role as primary production sites in coastal ecosystems but face threats by isoyake (the denudation of rocky shore seaweeds) and other environmental factors. To support effective conservation and enable early detection of seaweed bed decline, more efficient monitoring techniques are needed to address the limitations of conventional survey methods. In this study, we developed a technique to simultaneously detect four major seaweed and seagrass families (Laminariaceae, Sargassaceae, Zosteraceae, and Alariaceae) and bedrock (five categories in total) from underwater images using convolutional neural network-based transfer learning and fine-tuning. The developed method maintained high prediction accuracy across images taken under various environmental conditions, highlighting its potential for practical applications.</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

Development of a seaweed bed monitoring method using image analysis with convolutional neural networks

  • 期刊论文

  • 发表日期：

    2026年5月26日

查看全文
  • Yasuyuki Kanamoto

摘要

Seaweed beds play a crucial role as primary production sites in coastal ecosystems but face threats by isoyake (the denudation of rocky shore seaweeds) and other environmental factors. To support effective conservation and enable early detection of seaweed bed decline, more efficient monitoring techniques are needed to address the limitations of conventional survey methods. In this study, we developed a technique to simultaneously detect four major seaweed and seagrass families (Laminariaceae, Sargassaceae, Zosteraceae, and Alariaceae) and bedrock (five categories in total) from underwater images using convolutional neural network-based transfer learning and fine-tuning. The developed method maintained high prediction accuracy across images taken under various environmental conditions, highlighting its potential for practical applications.