Koval, Anton (Luleå University of Technology), Sharif Mansouri, Sina (Luleå University of Technology), Kanellakis, Christoforos (Luleå University of Technology), Nikolakopoulos, George (Luleå University of Technology)

Aerial Thermal Image Based Convolutional Neural Networks for Human Detection in SubT Environments

Scheduled for presentation during the Regular Session "UAS Applications I" (ThA4), Thursday, June 17, 2021, 10:30−10:50, Naoussa

2021 International Conference on Unmanned Aircraft Systems (ICUAS), June 15-18, 2021, Athens, Greece

This information is tentative and subject to change. Compiled on April 23, 2024

Abstract

This article proposes a novel strategy for detecting humans in harsh Sub-terranean (SubT) environments, with a thermal camera mounted on an aerial platform, based on the AlexNet Convolutional Neural Network (CNN). A transfer learning framework will be utilized for detecting the humans, where the aerial thermal images are fed to the trained network, which binary classifies them image content into two categories: a) human, and b) no human. Moreover, the AlexNet based framework is compared with two related popular CNN approaches as the GoogleNet and the Inception3Net. The efficacy of the proposed scheme has been experimentally evaluated through multiple data-sets, collected from a FLIR thermal camera during flights on an underground mining environment, fully demonstrating the performance and merits of the proposed module.