Deep learning-based acoustic event detection and localization for smart indoor security systems

Sediono, Wahju and Noor Fahmy, Norhezry Hakimie (2026) Deep learning-based acoustic event detection and localization for smart indoor security systems. In: 2026 10th International Conference on Mechatronics Engineering (ICOM), 8-9 September 2026, Kuala Lumpur.

PDF - Published Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Acoustic event detection has become an essential component of intelligent security systems due to its ability to identify abnormal events in environments where visual monitoring is limited or unavailable. Among various acoustic events, glass breaking is a critical indicator of potential intrusion and requires reliable detection with minimal false alarms. This paper presents an integrated Acoustic Event Detection and Localization (AEDL) framework for smart security applications, combining deep learning-based sound classification with Direction of Arrival (DOA) estimation. In this work, localization is defined as estimating the DOA of the detected sound source using a four-channel MEMS microphone array. A lightweight Convolutional Neural Network (CNN) is employed to classify glass-breaking events from log-Mel spectrogram features, while the Generalized Cross-Correlation with Phase Transform (GCCPHAT) algorithm is used to estimate the DOA. The proposed system is implemented on an ESP32-S3-based sensing platform for embedded audio acquisition with host-based real-time signal processing. Experimental results show that the CNN achieves an offline classification accuracy of 93.5%, while the integrated hardware prototype attains 82.5% accuracy under controlled indoor conditions. These results demonstrate the feasibility of integrating deep learning and array signal processing to provide a practical, low-cost solution for intelligent acoustic security monitoring in smart environments.

Item Type:	Proceeding Paper (Invited Papers)
Uncontrolled Keywords:	Acoustic event detection, environmental sound classification, glass-breaking detection, convolutional neural network, GCC-PHAT, direction of arrival estimation, microphone array, embedded acoustic sensing
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7800 Electronics. Computer engineering. Computer hardware. Photoelectronic devices > TK7885 Computer engineering
Kulliyyahs/Centres/Divisions/Institutes (Can select more than one option. Press CONTROL button):	Kulliyyah of Engineering > Department of Mechatronics Engineering Kulliyyah of Engineering
Depositing User:	Dr.-Ing. Wahju Sediono
Date Deposited:	27 Oct 2026 11:10
Last Update:	27 Oct 2026 11:10
Queue Number:	2026-10-Q5818
URI:	http://irep.iium.edu.my/id/eprint/132235
Indexed In:	Google Scholar

Actions (login required)

View Item