Deep Sensing of Urban Waterlogging

In the monsoon season, sudden flood events occur frequently in urban areas, which hamper social and economic activities may threaten infrastructure lives. The use of an efficient large-scale waterlogging sensing information system can provide valuable near real-time disaster to facilitate management enhance awareness general public alleviate losses during after disasters. Therefore, this study, a visual approach driven by deep neural networks communication technology was developed end-to-end mechanism realize event-location mapping. season Taiwan demonstrated, were predicted on island-wide scale. could sense approximately 2379 vision sources through internet video things framework transmit 5 min. proposed at national scale highly scalable alternative conventional methods.