Scalable Flood Level Trend Monitoring with Surveillance Cameras using a Deep Convolutional Neural Network

Abstract

Abstract. In many countries, urban flooding due to local, intense rainfall is expected to become more frequent because of climate change and urbanization. Cities trying to adapt to this growing risk are challenged by a chronic lack of surface flooding data that is needed for flood risk assessment and planning. In this work, we propose a new approach that exploits existing surveillance camera systems to provide qualitative flood level trend information at scale. The approach uses a deep convolutional neural network (DCNN) to detect floodwater in surveillance footage and a novel qualitative flood index (SOFI) as a proxy for water level fluctuations visible from a surveillance camera’s viewpoint. To demonstrate the approach, we trained the DCNN on 1281 flooding images collected from the Internet and applied it to six surveillance videos representing different flooding and lighting conditions. The SOFI signal obtained from the videos had on average a 75 % correlation to the actual water level fluctuation. By retraining the DCNN with a few frames from a given video, performance for that video is further increased to 85 %. The results suggest that the approach can be used with almost any surveillance footage without the need for on-site camera calibration, making it cheap, highly scalable, and retroactively applicable to archived surveillance footage.

Deep learning for the estimation of water-levels using river cameras

“…For example, without any type of metadata regarding the height of locations within the field-of-view of the camera, a first generic approach presented by Moy de Vitry et al (2019) proposes to use the percentage of water pixels in the image to observe the relative evolution of the water-level.…”

“…In Moy de Vitry et al (2019), the authors also proposed to use a deep semantic segmentation network trained from scratch to produce a generic algorithm for flood level trend monitoring. The biggest originality of this paper lies in their development of the SOFI index that corresponds to the percentage of pixels estimated as water pixels by the network in the image.…”

“…CC BY 4.0 License. networks are trained on small datasets composed of 100-300 training images (Lopez-Fuentes et al, 2017;Steccanella et al, 2018;Moy de Vitry et al, 2019) while more popular problems can be trained on datasets composed of more than 15000 images (e.g, Caesar et al (2018); Zhou et al (2017)). In many cases, using inductive TL approaches for the training of CNNs instead of training them from scratch, with randomly initialised weights, allows improvement in the network performance (Reyes et al, 2015;Sabatelli et al, 2018).…”

Water Level Recognition on Resource-Constrained Equipment using Pruned YOLOv5 and Template Matching

“…Consequently, more attention has been paid to non-contact water level measurement technology. For example, [3] proposes a deep convolutional neural network to detect floodwater in surveillance footage and proposes a simple but effective qualitative flood index to monitor flood level trend. Additionally, [4] uses a mobile app with virtual staff gauges to upload water level images taken by citizens at different periods.…”

Water Level Fluctuation Using Surveillance Camera

“…Therefore, this paper demonstrates the use of the hybrid technique for water segmentation tasks. For water level fluctuation, a flood index was introduced with the same concept as the one proposed by Moy de Vitry et al (2019). To evaluate the performance of the flood index, the results were compared with water level measured by sensor on-site.…”