Home > Research > Publications & Outputs > An Interpretable Deep Semantic Segmentation Met...

Research

Associated organisational units

Links

Text available via DOI:

View graph of relations

An Interpretable Deep Semantic Segmentation Method for Earth Observation.

Research output: Contribution to Journal/Magazine › Journal article › peer-review

Published

Standard

An Interpretable Deep Semantic Segmentation Method for Earth Observation. / Zhang, Ziyang; Angelov, Plamen; Soares, Eduardo et al.
In: arXiv, Vol. abs/2210.12820, 23.10.2022.

Research output: Contribution to Journal/Magazine › Journal article › peer-review

Harvard

Zhang, Z, Angelov, P, Soares, E, Longépé, N & Mathieu, P-P 2022, 'An Interpretable Deep Semantic Segmentation Method for Earth Observation.', arXiv, vol. abs/2210.12820. https://doi.org/10.48550/arXiv.2210.12820

APA

Zhang, Z., Angelov, P., Soares, E., Longépé, N., & Mathieu, P.-P. (2022). An Interpretable Deep Semantic Segmentation Method for Earth Observation. arXiv, abs/2210.12820. https://doi.org/10.48550/arXiv.2210.12820

Vancouver

Zhang Z, Angelov P, Soares E, Longépé N, Mathieu PP. An Interpretable Deep Semantic Segmentation Method for Earth Observation. arXiv. 2022 Oct 23;abs/2210.12820. doi: 10.48550/arXiv.2210.12820

Author

Zhang, Ziyang ; Angelov, Plamen ; Soares, Eduardo et al. / An Interpretable Deep Semantic Segmentation Method for Earth Observation. In: arXiv. 2022 ; Vol. abs/2210.12820.

Bibtex

@article{0e8afc07eb224a85a63652056c9abcbe,
title = "An Interpretable Deep Semantic Segmentation Method for Earth Observation.",
abstract = "Earth observation is fundamental for a range of human activities including flood response as it offers vital information to decision makers. Semantic segmentation plays a key role in mapping the raw hyper-spectral data coming from the satellites into a human understandable form assigning class labels to each pixel. In this paper, we introduce a prototype-based interpretable deep semantic segmentation (IDSS) method, which is highly accurate as well as interpretable. Its parameters are in orders of magnitude less than the number of parameters used by deep networks such as U-Net and are clearly interpretable by humans. The proposed here IDSS offers a transparent structure that allows users to inspect and audit the algorithm's decision. Results have demonstrated that IDSS could surpass other algorithms, including U-Net, in terms of IoU (Intersection over Union) total water and Recall total water. We used WorldFloods data set for our experiments and plan to use the semantic segmentation results combined with masks for permanent water to detect flood events.",
author = "Ziyang Zhang and Plamen Angelov and Eduardo Soares and Nicolas Long{\'e}p{\'e} and Pierre-Philippe Mathieu",
year = "2022",
month = oct,
day = "23",
doi = "10.48550/arXiv.2210.12820",
language = "English",
volume = "abs/2210.12820",
journal = "arXiv",
issn = "2331-8422",

}

RIS

TY - JOUR

T1 - An Interpretable Deep Semantic Segmentation Method for Earth Observation.

AU - Zhang, Ziyang

AU - Angelov, Plamen

AU - Soares, Eduardo

AU - Longépé, Nicolas

AU - Mathieu, Pierre-Philippe

PY - 2022/10/23

Y1 - 2022/10/23

N2 - Earth observation is fundamental for a range of human activities including flood response as it offers vital information to decision makers. Semantic segmentation plays a key role in mapping the raw hyper-spectral data coming from the satellites into a human understandable form assigning class labels to each pixel. In this paper, we introduce a prototype-based interpretable deep semantic segmentation (IDSS) method, which is highly accurate as well as interpretable. Its parameters are in orders of magnitude less than the number of parameters used by deep networks such as U-Net and are clearly interpretable by humans. The proposed here IDSS offers a transparent structure that allows users to inspect and audit the algorithm's decision. Results have demonstrated that IDSS could surpass other algorithms, including U-Net, in terms of IoU (Intersection over Union) total water and Recall total water. We used WorldFloods data set for our experiments and plan to use the semantic segmentation results combined with masks for permanent water to detect flood events.

AB - Earth observation is fundamental for a range of human activities including flood response as it offers vital information to decision makers. Semantic segmentation plays a key role in mapping the raw hyper-spectral data coming from the satellites into a human understandable form assigning class labels to each pixel. In this paper, we introduce a prototype-based interpretable deep semantic segmentation (IDSS) method, which is highly accurate as well as interpretable. Its parameters are in orders of magnitude less than the number of parameters used by deep networks such as U-Net and are clearly interpretable by humans. The proposed here IDSS offers a transparent structure that allows users to inspect and audit the algorithm's decision. Results have demonstrated that IDSS could surpass other algorithms, including U-Net, in terms of IoU (Intersection over Union) total water and Recall total water. We used WorldFloods data set for our experiments and plan to use the semantic segmentation results combined with masks for permanent water to detect flood events.

U2 - 10.48550/arXiv.2210.12820

DO - 10.48550/arXiv.2210.12820

M3 - Journal article

VL - abs/2210.12820

JO - arXiv

JF - arXiv

SN - 2331-8422

ER -