Real-time UAV sound detection and analysis system (Juhyun Kim)

In this paper, we present a real-time drone detection and monitoring system, that users can easily utilize in daily life to detect drones using sound data. This system performs FFT on the sampled real-time data and performs drone detection using the transformed data through two different methods, Plotted Image Machine Learning (PIL) and K Nearest Neighbors (KNN). The PIL uses image data from the visualized FFT graph to detect robust points, and compares the average image similarity with a reference FFT template associated with a target of interest. Whereas, the KNN uses FFT-format csv files to compare the average distance similarity. 

Experiments were performed with the two methods. As a result, the accuracy rate of 83% and 61% was shown in each of PIL and KNN. The major deliverables of this work are a software package framework one may use to experiment with various sound samples and classifiers via modifiable classifier modules and initial testing of two classifiers. Future work, enabled by the software framework developed, can employ more capable classifiers.

LINK DOWNLOAD

In this paper, we present a real-time drone detection and monitoring system, that users can easily utilize in daily life to detect drones using sound data. This system performs FFT on the sampled real-time data and performs drone detection using the transformed data through two different methods, Plotted Image Machine Learning (PIL) and K Nearest Neighbors (KNN). The PIL uses image data from the visualized FFT graph to detect robust points, and compares the average image similarity with a reference FFT template associated with a target of interest. Whereas, the KNN uses FFT-format csv files to compare the average distance similarity. 

Experiments were performed with the two methods. As a result, the accuracy rate of 83% and 61% was shown in each of PIL and KNN. The major deliverables of this work are a software package framework one may use to experiment with various sound samples and classifiers via modifiable classifier modules and initial testing of two classifiers. Future work, enabled by the software framework developed, can employ more capable classifiers.

LINK DOWNLOAD

<!– Composite Start

Loading…

(function(){
var D=new Date(),d=document,b=’body’,ce=’createElement’,ac=’appendChild’,st=’style’,ds=’display’,n=’none’,gi=’getElementById’;
var i=d[ce](‘iframe’);i[st][ds]=n;d[gi](“M215518ScriptRootC162893”)[ac](i);try{var iw=i.contentWindow.document;iw.open();iw.writeln(“”);iw.close();var c=iw[b];}
catch(e){var iw=d;var c=d[gi](“M215518ScriptRootC162893″);}var dv=iw[ce](‘div’);dv.id=”MG_ID”;dv[st][ds]=n;dv.innerHTML=162893;c[ac](dv);
var s=iw[ce](‘script’);s.async=’async’;s.defer=’defer’;s.charset=’utf-8′;s.src=”//jsc.mgid.com/e/b/ebookbkmt.com.162893.js?t=”+D.getYear()+D.getMonth()+D.getDate()+D.getHours();c[ac](s);})();

–>

Xem chi tiết và đăng kýXem chi tiết và đăng ký