RipNet: A Lightweight One-Class Deep Neural Network for the Identification of RIP Currents

Abstract

Rip or rip current is a strong, localized and narrow current of water flowing away from shore through the surf zone, cutting through the lines of breaking ocean waves. There are hundreds of deaths due to drowning and 85% of rescues missions on beaches are due to rip currents. Although, there are rare drowning between flags, however, we can not put and monitor enough flags. Automated rip current identification can help to monitor the coast however there are several challenges involved in development of automated rip current identification. In this work, we present an automated rip current identification based on fully convolutional autoencoder. The proposed framework is able to reconstruct the positive RIP currents images with minimal root mean square error (RMSE). Evaluation results on Rip currents dataset showed an increase in accuracy, specificity and sensitivity to 99.40% 99.134%, and 93.427% respectively in comparison to state of the art methods. © 2020, Springer Nature Switzerland AG.