BROWSE
Detail
Conference Paper
Intersection Flood Detection Using a Grid-Based Two-Stage Classifier
Cited - time in 
Share 
Share
- Authors
- Jang Woon Baek, Jinhong Kim, Kwangju Kim, Yun Won Choi
- Issue Date
- 2026-02
- Citation
- International Conference on Artificial Intelligence in Information and Communication (ICAIIC) 2026, pp.1393-1395
- Publisher
- IEEE
- Language
- English
- Type
- Conference Paper
- Abstract
- Frequent urban flooding caused by extreme weather events has emerged as a serious threat to traffic safety and infrastructure. Traditional flood detection systems, such as those based on IoT sensors or hydrological simulations, are limited by high installation costs and poor scalability. To address these issues, this study proposes an intersection flood detection system based on a grid-based two-stage classifier using CCTV images. The first stage classifier determines whether the road surface is dry or wet. The second stage classifier determines whether a road is flooded or not. To determine whether a road is normal, partially flooded, or completely flooded, the image is divided into grids and flooding is determined for each grid. This allows us to identify which grids are flooded and determine the extent of the flooding. The proposed method enables fast and reliable detection even in complex intersection environments. Experimental results demonstrate high detection accuracy and robustness against lighting and occlusion, confirming its potential for real-time deployment in smart city infrastructure.
- Keyword
- Flood detection, Two-Stage Classifier, Intersection Surveillance
- KSP Keywords
- CCTV images, Detection accuracy, Extreme weather events, First stage, Flood detection system, Grid-based, Intrusion Detection Systems(IDS), IoT sensors, Smart City infrastructure, Traffic Safety, Urban flooding