Modernized Training of U-Net for Aerial Semantic Segmentation

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F49777513%3A23520%2F24%3A43973190" target="_blank" >RIV/49777513:23520/24:43973190 - isvavai.cz</a>

Výsledek na webu

<a href="https://ieeexplore.ieee.org/document/10495686" target="_blank" >https://ieeexplore.ieee.org/document/10495686</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/WACVW60836.2024.00091" target="_blank" >10.1109/WACVW60836.2024.00091</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Modernized Training of U-Net for Aerial Semantic Segmentation

Popis výsledku v původním jazyce

In this paper, we propose an improved training protocol of U-Net architecture for the semantic segmentation of aerial images. We test our approach on the challenging FLAIR #2 dataset. We present an extensive ablation study on the influence of different approach components on the overall performance. The ablation study includes a comparison of different model backbones, image augmentations, learning rate schedulers, loss functions, and training procedures. We additionally propose a two-stage training procedure and evaluate different options for the model ensemble. Based on the results we design the final setup of the model training protocol. This final setup decreases the relative error by approximately 18% and achieves mIoU equal to 0.641, which is a new state-of-the-art result. Our code is available at: https://github.com/strakaj/U-Net-for-remote-sensing

Název v anglickém jazyce

Modernized Training of U-Net for Aerial Semantic Segmentation

Popis výsledku anglicky

In this paper, we propose an improved training protocol of U-Net architecture for the semantic segmentation of aerial images. We test our approach on the challenging FLAIR #2 dataset. We present an extensive ablation study on the influence of different approach components on the overall performance. The ablation study includes a comparison of different model backbones, image augmentations, learning rate schedulers, loss functions, and training procedures. We additionally propose a two-stage training procedure and evaluate different options for the model ensemble. Based on the results we design the final setup of the model training protocol. This final setup decreases the relative error by approximately 18% and achieves mIoU equal to 0.641, which is a new state-of-the-art result. Our code is available at: https://github.com/strakaj/U-Net-for-remote-sensing

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

20205 - Automation and control systems

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

2024 IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW)

ISBN

979-8-3503-7028-7

ISSN

2572-4398

e-ISSN

2690-621X

Počet stran výsledku

9

Strana od-do

785-793

Název nakladatele

Institute of Electrical and Electronics Engineers Inc.

Místo vydání

Piscataway

Místo konání akce

Waikoloa

Datum konání akce

1. 1. 2024

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

001223022200092