@article{ART003280440},
author={Chanhee Kwak and Seong-Min Lee},
title={A CNN-based Water Pipe Leakage Classification Model Using Frequency-domain Feature Learning},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2025},
volume={30},
number={12},
pages={131-137}
TY - JOUR
AU - Chanhee Kwak
AU - Seong-Min Lee
TI - A CNN-based Water Pipe Leakage Classification Model Using Frequency-domain Feature Learning
JO - Journal of The Korea Society of Computer and Information
PY - 2025
VL - 30
IS - 12
PB - The Korean Society Of Computer And Information
SP - 131
EP - 137
SN - 1598-849X
AB - Leakage from aging infrastructure causes significant water losses and economic costs. Conventional acoustic detection methods are limited by their dependence on expert experience and susceptibility to external noise. While sensor-based data and machine learning methods have been introduced to address these issues, existing models treating frequency spectra as independent features suffer from class imbalance and limited capacity to capture global patterns, resulting in poor generalization. In this study, we design a hybrid architecture that combines 1D-CNN and 2D-CNN and validate its performance using the AI-Hub leak detection dataset. The proposed model leverages 1D-CNN to extract local patterns along the frequency axis and 2D-CNN to capture inter-channel interactions, thereby overcoming the shortcomings of previous approaches. Experimental results demonstrate that the hybrid architecture achieves an accuracy of 0.9587, a Macro F1 score of 0.9487, and an MCC of 0.9456, confirming its potential applicability to a wide range of frequency-based sensor data analysis tasks.
KW - Water pipeline;Leak detection;Frequency spectrum;CNN;Sensor data
DO -
UR -
ER -
Chanhee Kwak and Seong-Min Lee. (2025). A CNN-based Water Pipe Leakage Classification Model Using Frequency-domain Feature Learning. Journal of The Korea Society of Computer and Information, 30(12), 131-137.
Chanhee Kwak and Seong-Min Lee. 2025, "A CNN-based Water Pipe Leakage Classification Model Using Frequency-domain Feature Learning", Journal of The Korea Society of Computer and Information, vol.30, no.12 pp.131-137.
Chanhee Kwak, Seong-Min Lee "A CNN-based Water Pipe Leakage Classification Model Using Frequency-domain Feature Learning" Journal of The Korea Society of Computer and Information 30.12 pp.131-137 (2025) : 131.
Chanhee Kwak, Seong-Min Lee. A CNN-based Water Pipe Leakage Classification Model Using Frequency-domain Feature Learning. 2025; 30(12), 131-137.
Chanhee Kwak and Seong-Min Lee. "A CNN-based Water Pipe Leakage Classification Model Using Frequency-domain Feature Learning" Journal of The Korea Society of Computer and Information 30, no.12 (2025) : 131-137.
Chanhee Kwak; Seong-Min Lee. A CNN-based Water Pipe Leakage Classification Model Using Frequency-domain Feature Learning. Journal of The Korea Society of Computer and Information, 30(12), 131-137.
Chanhee Kwak; Seong-Min Lee. A CNN-based Water Pipe Leakage Classification Model Using Frequency-domain Feature Learning. Journal of The Korea Society of Computer and Information. 2025; 30(12) 131-137.
Chanhee Kwak, Seong-Min Lee. A CNN-based Water Pipe Leakage Classification Model Using Frequency-domain Feature Learning. 2025; 30(12), 131-137.
Chanhee Kwak and Seong-Min Lee. "A CNN-based Water Pipe Leakage Classification Model Using Frequency-domain Feature Learning" Journal of The Korea Society of Computer and Information 30, no.12 (2025) : 131-137.
