Gunshot Detection Using Convolutional Neural Networks

Abstract

The main paper deals with the analysis of the methods of signal processing and events recognition in the audio signal and the implementation of the selected method in real use. Recognized events are gunshots mixed with a background sound such as traffic noise, human voice, animal sounds and other forms of environmental sounds. The proposed algorithm adapted for explosion detection can be used as part of a security system for monitoring depots or places dedicated to storing dangerous materials. For events classification and class recognition, the freely available machine learning frameworks TensorFlow and Keras are used.