Segmentation of water bodies in Chennai is vital, as stagnant waters and flooding due to heavy rain and blockage in drainage are recurrent issues. Being a coastal city, Chennai has its share of difficulties with water resource management during monsoon seasons. Proper identification and segmentation of water bodies lessen the impacts of flooding and, at the same time, contribute to urban planning and water management. The results of this study in water-body segmentation are of great importance, presenting a comprehensive understanding of Chennai’s water systems and revealing the current level of water management in the city. It employs Geographic Information System technology to identify and locate the water systems found in Chennai. A high-resolution Digital Elevation Model of the elevations in meters above sea level was derived from the Shuttle Radar Topographic Mission. This dataset comprises those water bodies verified through satellite imagery and the DEM data in the intersection. The SegFormer model, which promises to be a game changer in water body segmentation, was then applied to segment the water bodies. This provides efficient identification and extraction of the water bodies required to manage further Chennai’s urban water infrastructure and flood mitigation measures. Using DEM data, SegFormer achieved a training accuracy of 90.58% and a testing accuracy of 94.77%.

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

Segmentation of Water Bodies in Chennai Using SegFormer

  • Ashutosh Satapathy,
  • Thottempudi Kokila,
  • Polukonda Kalyani

摘要

Segmentation of water bodies in Chennai is vital, as stagnant waters and flooding due to heavy rain and blockage in drainage are recurrent issues. Being a coastal city, Chennai has its share of difficulties with water resource management during monsoon seasons. Proper identification and segmentation of water bodies lessen the impacts of flooding and, at the same time, contribute to urban planning and water management. The results of this study in water-body segmentation are of great importance, presenting a comprehensive understanding of Chennai’s water systems and revealing the current level of water management in the city. It employs Geographic Information System technology to identify and locate the water systems found in Chennai. A high-resolution Digital Elevation Model of the elevations in meters above sea level was derived from the Shuttle Radar Topographic Mission. This dataset comprises those water bodies verified through satellite imagery and the DEM data in the intersection. The SegFormer model, which promises to be a game changer in water body segmentation, was then applied to segment the water bodies. This provides efficient identification and extraction of the water bodies required to manage further Chennai’s urban water infrastructure and flood mitigation measures. Using DEM data, SegFormer achieved a training accuracy of 90.58% and a testing accuracy of 94.77%.