Evaluation of a YOLOv8-Based Method for Detecting Unauthorized Airstrips in the Amazon Rainforest using SAR Imagery

Leandro da Silva Gomes, Elcio Hideiti Shiguemori, Tahisa Neitzel Kuck, Dimas Irion Alves

DOI: 10.14209/sbrt.2024.1571036260

Evento: XLII Simpósio Brasileiro de Telecomunicações e Processamento de Sinais (SBrT2024)

Keywords: SAR images illegal airstrips Amazon Rainforest Sentinel-1

Abstract

This study presents the application of a method to detect unauthorized airstrips in the Brazilian Amazon using synthetic aperture radar (SAR) images from the Sentinel-1 satellite. There are illegal activities, such as mining and drug trafficking, which use these airstrips for their operations. Furthermore, the area is known for its dense vegetation. A YOLOv8 model was trained on a database comprising 646 training, 277 validation, and 109 generalization testing images. Results demonstrate the model's effectiveness in identifying unauthorized airstrips, highlighting its potential for monitoring remote regions. The central experiment correctly identified 39 out of 109 landing strips.

Download