‘Real-time UAV Sound Detection and Analysis’ IEEE Sensors Applications Symposium accepted paper with Machine Learning Framework

Keywords: Machine learning; Audio categorization; Drone classification; K-NN; UAV categorization; UAV analysis; Structural similarity index

Juhyun Kim, Cheonbok Park, Jinwoo Ahn, Youlim Ko, Junghyun Park, John Gallagher.

‘Real-time UAV Sound Detection and Analysis’ is a 2017 IEEE Sensors Applications Symposium (SAS) accepted paper(publication in the proceedings) based on Drone detecting Machine Learning Framework.

Full paper was presented in the 2017 IEEE SAS by the first author & oral presenter Juhyun Kim(drexly), New Jersey, USA, 2017.03.08

• Devised 2 machine learning algoritms and developed 1-Server:6-Clients network interacting UAV air defense GUI system framework by Python which can locate the UAV’s appearance by analyzing and machine learning its features (with Realtime Fast Fourier transformed Spectogram) through real-time transformed data inputs of UAV frequency from motor sounds

The full text paper was peer-reviewed at the direction of IEEE Instrumentation and Measurement Society prior to the acceptance and publication. http://ieeexplore.ieee.org/document/7894058/

• The paper proposes a novel theme surrounding use of Artificial Intelligence by employing learning/training of Artificial Neural Networks to predict presence/absence of object of interest, Drone in this case. The software framework put together is impressive and exhibits thoughtful process of practical experimentation.
  Since the paper attempts to highlight better use of inexpensive sensing technology, more details on the proposed placement methods of the inexpensive microphone sensor could be valuable to the reader, especially someone interested in real-time monitoring as mentioned in the paper. In addition, in a simplistic view, the central idea is to process the measured audio signal, convert it into a data-set, and use the data-set to train the ANN. The aspects surrounding ANN training and prediction is not obvious from the given description.
  Overall, a good framework for future experimentation.

• Authors propose real-time detection and monitoring by low cost system using inexpensive microphones and devices.
  I strongly recommend this paper to SAS 2017.

System Demo Video(Click Below)

Presentation in 2017 IEEE SAS(Just a photo)

Futureworks

Ongoing SCI-TIM Applying by granted as “IEEE TIM 2017 - SAS 2017 Special issue”

Marquis who’s who nomination