Dr. Kumpee Teeravech

Rambai Barni Rajabhat University, Thailand

Ship Types Classification using DEMON and LOFAR Techniques


Kumpee Teeravech is now a head of Geoinformatics Program, Faculty of Computer Science and Information Technology (CSIT), Rambhai Barni Rajabhat Universit (RBRU), Thailand. He is also a member of the digital innovation and research committee at CSIT. He received his bachelor’s degree in computer science from Rambhai Barni Rajabhat Institute (RBRI), the former name of RBRU. Then, he obtained his M.Sc. (Royal Thai Government scholarship) and Ph.D. (Japan scholarship) in Remote Sensing and Geographic Information Systems from Asian Institute of Technology (AIT), Thailand. During his Ph.D. study in AIT, he spent some of his time as a research assistant at CHUBU university, Japan. Before joining RBRU in 2004, he worked as a freelance programmer for more than 10 years. He was one of the co-founders of a startup company which is now offering several agricultural technologies and solutions to farmers in Thailand. In 2017, he and his colleagues won two best awards from the IoT Data Alchemist Hackathon which was held during 16 – 23 November, 2017, at Jeju, Korea. More recently, Dr. Kumpee and his colleagues won the GISTDA award from the RPD Challenge 2022. He is currently serving as a reviewer for the Defence Technology Academic Journal (DTAJ).

His research interests include digital image processing, digital photogrammetry, structure from motion, remote sensing, computer vision, and artificial intelligence. He is also interested in unmanned aerial vehicle and unmanned underwater vehicle. Dr. Kumpee sometimes spend his holiday building fixed-wing unmanned aerial vehicles from scratch.


This talk focuses on the concept of analyzing and exacting underwater audio features for ship/vessel sound classification problem. The basics of signal processing, i.e., filtering and resampling, are presented. Then, the DEMON (Detecting Envelope Modulation on Noise) and LOFAR (Low Frequency Analysis and Recording) techniques are introduced and demonstrated. The implementation of the deep learning on a STM32 microcontroller is also presented. The length of the talk would be 30 minutes approximately.