Scholarlink

<p>Litter is a significant contaminant in the world’s oceans. Mismanaged land-based garbage reaches the marine environment via rivers, creeks, and other water bodies. In this study, targeted debris is detected on the beaches and creeks of Chennai using various satellite sensors with specific spectral bands. Additionally, the effectiveness of drone imagery is assessed in detecting macro litter along the coastal area. Pixel-based semantic segmentation deep learning techniques are applied to differentiate among various litter classes. The results reveal an estimation of litter abundance and the spatial distribution of debris on the beaches, as shown by field-collected ground truth data and segregation statistics. The outcomes demonstrate natural and artificial litter organised into eight segmented classes (styrofoam, chappals, wood, water bottles, wrappers, glass bottles, vegetal waste, and other plastic litter), with the mean Intersection over Union (mIoU) for the Validation set being 0.0834 and for the Test set being 0.0870. The model could not confidently detect or distinguish similar textures, such as sand. This study highlights the urgent need for sustainable waste management practices and emphasizes their critical national importance, particularly in facilitating immediate decision-making during extreme weather conditions.</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

Monitoring of Marine Floating Debris and Beach Litter Using Satellite and Drones: A Synergistic Approach on Policy & Decision Making

  • 期刊论文

  • 发表日期：

    2025年11月11日

查看全文
  • P. Thanabalan,
  • Kharatmole Gayathrri,
  • Mitsuko Hidaka,
  • Daisuke Matsuoka,
  • Pravakar Mishra,
  • Tune Usha,
  • Heidi Dierssen,
  • Sisir Kumar Dash,
  • Shambanagouda R. Marigoudar

摘要

Litter is a significant contaminant in the world’s oceans. Mismanaged land-based garbage reaches the marine environment via rivers, creeks, and other water bodies. In this study, targeted debris is detected on the beaches and creeks of Chennai using various satellite sensors with specific spectral bands. Additionally, the effectiveness of drone imagery is assessed in detecting macro litter along the coastal area. Pixel-based semantic segmentation deep learning techniques are applied to differentiate among various litter classes. The results reveal an estimation of litter abundance and the spatial distribution of debris on the beaches, as shown by field-collected ground truth data and segregation statistics. The outcomes demonstrate natural and artificial litter organised into eight segmented classes (styrofoam, chappals, wood, water bottles, wrappers, glass bottles, vegetal waste, and other plastic litter), with the mean Intersection over Union (mIoU) for the Validation set being 0.0834 and for the Test set being 0.0870. The model could not confidently detect or distinguish similar textures, such as sand. This study highlights the urgent need for sustainable waste management practices and emphasizes their critical national importance, particularly in facilitating immediate decision-making during extreme weather conditions.