Optimizing Spectral Inputs for Deep Learning-Based Vegetation Destruction Detection Using Sentinel-2 Imagery
Remote Sensing · 13.07.2026 · 18 (14) · 2333
DOI: 10.3390/rs18142333 ↗Kurz gefasst
Fachartikel in der Zeitschrift Remote Sensing, online erschienen am 13.07.2026. Verfasst von Chuanwu Zhao und 3 weiteren Autor:innen. Die Arbeit nennt 66 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: China Postdoctoral Science Foundation, the Postdoctoral Research Funding of Hangzhou International Innovation Institute of Beihang University.
Abstract (Originalsprache)
Deep learning models have been widely used for vegetation destruction detection, but their performance depends on input features in addition to sample size and model architecture. Using Sentinel-2 imagery, we constructed multiple feature combinations from visible, near-infrared, and short-wave infrared bands and several spectral indices, and evaluated them with nine deep learning models. The results show that adding extra bands to RGB input improved most models, with RGBS yielding the most robust performance improvement among the tested band combinations, consistently outperforming the RGB baseline across the majority of deep learning architectures. Spectral indices further improved most models, especially BVDI, NBR, and SVI. Some single-index or two-index inputs achieved accuracy comparable to or higher than that of RGB input, indicating their potential for information compression and band substitution. However, model responses varied, and more input channels did not guarantee stable gains. Accuracy–efficiency analysis showed that RGBS, RGB + BVDI, RGB + NBR, and selected RGBNS + index combinations achieved favorable performance at low computational cost. Overall, optimized spectral inputs improve vegetation destruction detection, but their selection should balance model architecture, accuracy, and computational cost.
Förderung
- China Postdoctoral Science Foundation
- the Postdoctoral Research Funding of Hangzhou International Innovation Institute of Beihang University
Zitieren
APA
Zhao, C., Zhao, H., Pan, Y., & Jing, G. (2026). Optimizing Spectral Inputs for Deep Learning-Based Vegetation Destruction Detection Using Sentinel-2 Imagery. Remote Sensing, 18(14), 2333. https://doi.org/10.3390/rs18142333
Vancouver
Zhao C, Zhao H, Pan Y, Jing G. Optimizing Spectral Inputs for Deep Learning-Based Vegetation Destruction Detection Using Sentinel-2 Imagery. Remote Sensing. 2026;18(14):2333. doi:10.3390/rs18142333
BibTeX
@article{Zhao2026,
author = {Zhao, Chuanwu and Zhao, Huijie and Pan, Yaozhong and Jing, Guifei},
title = {Optimizing Spectral Inputs for Deep Learning-Based Vegetation Destruction Detection Using Sentinel-2 Imagery},
journal = {Remote Sensing},
year = {2026},
volume = {18},
number = {14},
pages = {2333},
doi = {10.3390/rs18142333}
}
EUVIMED