Sign In

Communications of the ACM

ACM Careers

Researchers Train Model to Distinguish Gunshots From Similar Sounds

researcher and microphones recording sounds at an outdoor park

Researchers recorded gunshot-like sounds in various locations, including an outdoor park.

Credit: Florida Atlantic University

In an experimental study, researchers from Florida Atlantic University focused on addressing the reliability of gunshot detector systems and their ability to distinguish gunfire from similar sounds such as plastic bag explosions.

The researchers created a dataset comprised of audio recordings of plastic bag explosions collected over a variety of environments and conditions. They also developed a classification algorithm based on a convolutional neural network (CNN) to illustrate the relevance of the collected data. The data was then used, together with a gunshot sound dataset, to train a classification model based on a CNN to differentiate gunshot events from non-life-threatening events. 

Results are described in "Data Collection, Modeling, and Classification for Gunshot and Gunshot-like Audio Events: A Case Study," published in the journal Sensors. The model injected with plastic bag popping sounds performed well in distinguishing actual gunshot sounds from plastic bag sounds.

"We used different environments to give the machine learning algorithm a better perception sense of the differentiation of the closely related sounds," says Hanqi Zhuang, professor and chair in FAU's College of Engineering and Computer Science.

From Florida Atlantic University
View Full Article



No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account