Paper Contents
Abstract
Forests play an important role in protecting the environment and fighting global warming. Unfortunately, they are diminished by human intervention such as logging, fires, etc. This paper proposes and evaluates a framework to automatically detect illegal logging by classifying sound events. We develop small devices with ultra-low power consumption, embedded microcontrollers for edge processing and long-range wireless communication to cover large areas of the forest of . To reduce energy consumption and resource consumption for efficient and ubiquitous detection of illegal, efficient and accurate deforestation Audio classification based on convolution neural networks specifically designed for resource-constrained wireless edge devices. Compared to previous work, the proposed system detects a broader range of deforestation-related threats through a distributed and ubiquitous edge computing technique. Various pre-processing techniques were evaluated, with a focus on trade-offs between classification accuracy and computational resources, memory, and power consumption. In addition, experimental long-distance communication tests were carried out in real environments. Data from experimental results show that the proposed solution can detect and report felling events for efficient and cost-effective forest monitoring via intelligent IoT with 85% accuracy.
Copyright
Copyright © 2023 ANIL KUMAR KHANDALE. This is an open access article distributed under the Creative Commons Attribution License.
