抄録
Removing riverine floating debris is an essential environmental conservation practice to mitigate flood risks and reduce water pollution.This floating debris often accumulates around the upstream side of river weirs. While much of this riverine accumulated debris has traditionally been managed through regular manual patrols and removal, an automated detection system would significantly reduce the burden of these manual tasks. This paper reports on a deep learning-based method for the automated detection of accumulated floating debris around a river weir in Asuwa River. We constructed a deep learning model using a Convolutional Neural Network (CNN) trained on images captured by fixed cameras monitoring the river weir. Through repeated training of CNN and detection experiments, the CNN model achieved an accuracy of 0.86, a precision of 0.84, a recall of 0.90, and an F-measure of 0.87. This demonstrates the model's high effectiveness in identifying debris accumulation. The proposed system, utilizing fixed-camera monitoring, enables real-time detection, which can significantly reduce the need for manual patrols and contribute to more efficient river management.
https://ror.org/01n09qc98 
