Real-time and On the Edge Deforestation Detection

Abstract

Deforestation is one of the most challenging issue that we are dealing with as of today. The effects of complete deforestation have started to show up so now governments and authorities have started to take firm measures but unfortunately, illegal deforestation has still been going on. Real-time and on-the-edge Deforestation Detection is the need of the hour to control the unabated deforestation activities which are threatening the plant earth. For this purpose, Machine learning is an apt candidate to process audio signals in which deforestation related activities are to be detected in real-time. An embedded system, specifically designed for this purpose, is proposed which shall record sound, perform analysis and classify if currently there is any deforestation activity going on nearby. The developed model achieved more than 95% accuracy and was successfully deployed of Raspberry Pi boards