Acoustic Monitoring System with AI Threat Detection System for Forest Protection
Bhattarapong Somwong, Kritsana Kumphet, Wansuree Massagram · 2023
This paper presents an audio classification system specifically created for Portenta H7, an Arduino-based microcontroller. The proposed model utilizes Edge Impulse AI platform, which allows the creation of accurate and efficient classification models optimized for embedded systems. To evaluate the system performance, a set of experiments was conducted on a dataset of audio samples from four classes: chainsaw, handsaw, gunshot, and laugh - each depicted sounds involving illegal logging and poaching threat in the forests. The results demonstrate that the proposed approach achieved high accuracy for gunshot, satisfying accuracy for chainsaw and laugh, and unacceptable accuracy for handsaw from our satellite-enabled system. The proposed system also has potential applications in forest protection as well as various domains, such as smart homes, security systems, and healthcare, where accurate audio classification can enable intelligent decision-making.