Multi-Scale Hierarchical Attention Ensemble Network for Fine-Grained Riverine Waste Segmentation Using UAV Multispectral Imagery
Hydrology · 18.08.2026 · 13 (8) · 221
DOI: 10.3390/hydrology13080221 ↗Kurz gefasst
Fachartikel in der Zeitschrift Hydrology, online erschienen am 18.08.2026. Verfasst von Yohanes Fridolin Hestrio und 13 weiteren Autor:innen. Die Arbeit nennt 58 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Universitas Negeri Malang (UM) PKS RIIM KONEKSI, Australian KONEKSI Collaborative Research.
Abstract (Originalsprache)
Riverine plastic waste is difficult to detect and map accurately because debris ranges from small items to large floating clusters, and tropical rivers present challenging conditions, such as murky water, floating vegetation, and variable lighting. This study develops and benchmarks a deep learning method for pixel-level, multi-scale mapping of riverine waste from five-band UAV multispectral imagery. We deployed a drone equipped with a five-band multispectral sensor over the Brantas River in Surabaya, East Java, Indonesia, and introduce the Multi-Scale Hierarchical Attention Ensemble (MHAE), which combines three backbone networks across three image resolutions through learned scale- and backbone-attention weighting. MHAE was evaluated against 12 CNN-, transformer-, state-space-, and traditional-machine-learning-based baselines (including Random Forest, U-Net, UNet++, and DeepLabV3+) on the accompanying BrantasRiverWaste-UAV dataset (882 image tiles from a single-site, single-season orthomosaic covering approximately 0.54 km2 of river surface, labelled as water, land, organic waste, or inorganic waste), with pairwise comparisons assessed using Wilcoxon signed-rank tests with Bonferroni correction. MHAE achieved the highest pixel-level waste detection rate among the 12 evaluated models (86.76%), with a mean intersection-over-union of 78.33% (third-highest, behind UNet++ and U-Net). This work provides an initial, single-site benchmark and reference architecture for near-real-time riverine waste monitoring, and introduces the BrantasRiverWaste-UAV as, to our knowledge, one of the first five-band multispectral UAV datasets for tropical riverine waste mapping. Balanced sampling and multi-scale attention fusion can substantially improve waste-pixel detection under severe class imbalance; because the benchmark derives from a single site and season, future work should extend evaluation across additional seasons and river systems before the approach is generalised operationally.
Förderung
- Universitas Negeri Malang (UM) PKS RIIM KONEKSI
- Australian KONEKSI Collaborative Research
Zitieren
APA
Hestrio, Y. F., Nugroho, G., Karolinoerita, V., Candra, D. S., Susantoro, T. M., Sunarmodo, W., Wiwoho, B. S., Astuti, I. S., Sari, S. H. J., Sutapa, I. N., Panjaitan, T. W. S., Setyorini, D., Bulovic, N., & McIntyre, N. (2026). Multi-Scale Hierarchical Attention Ensemble Network for Fine-Grained Riverine Waste Segmentation Using UAV Multispectral Imagery. Hydrology, 13(8), 221. https://doi.org/10.3390/hydrology13080221
Vancouver
Hestrio YF, Nugroho G, Karolinoerita V, Candra DS, Susantoro TM, Sunarmodo W, et al. Multi-Scale Hierarchical Attention Ensemble Network for Fine-Grained Riverine Waste Segmentation Using UAV Multispectral Imagery. Hydrology. 2026;13(8):221. doi:10.3390/hydrology13080221
BibTeX
@article{Hestrio2026,
author = {Hestrio, Yohanes Fridolin and Nugroho, Gatot and Karolinoerita, Vicca and Candra, Danang Surya and Susantoro, Tri Muji and Sunarmodo, Wismu and Wiwoho, Bagus Setiabudi and Astuti, Ike Sari and Sari, Syarifah Hikmah Julinda and Sutapa, I Nyoman and Panjaitan, Togar Wiliater Soaloon and Setyorini, Daru and Bulovic, Nevenka and McIntyre, Neil},
title = {Multi-Scale Hierarchical Attention Ensemble Network for Fine-Grained Riverine Waste Segmentation Using UAV Multispectral Imagery},
journal = {Hydrology},
year = {2026},
volume = {13},
number = {8},
pages = {221},
doi = {10.3390/hydrology13080221}
}
EUVIMED