Comparison of Plain and Dense Skip Connections on U-Net Architecture for Change Detection

DOI

10.2991/978-94-6463-094-7_42How to use a DOI?

Keywords

U-Net; Skip Connection; Change Detection; CNN

Abstract

In recent years, identifying changes in multitemporal images in terms of land use and land cover is significant in a variety of applications including urban planning. CNN architectures are one of the most extensively utilised methods for change detection. The aim of this research is to investigate two types of skip connections that may be incorporated into CNN architecture to determine if they can improve the effectiveness of change detection during the CNN learning process. In this paper, we adopt the U-Net architecture to train the change detection model. We also modify the U-Net skip connection's path to include the dense skip connection and compare the modified U-Net with the original U-Net, which uses the plain skip connection. We also test the trained model with our collected local dataset in Cyberjaya to see how well it can anticipate changes in our location. The results of this study show that a U-Net with dense skip connections produces the best results and optimises change detection. It will help researchers understand how important the skip connection is to the model's performance.

Copyright

© 2022 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Zamfirdaus Saberi
AU  - Noramiza Hashim
PY  - 2022
DA  - 2022/12/27
TI  - Comparison of Plain and Dense Skip Connections on U-Net Architecture for Change Detection
BT  - Proceedings of the International Conference on Computer, Information Technology and Intelligent Computing (CITIC 2022)
PB  - Atlantis Press
SP  - 524
EP  - 532
SN  - 2589-4900
UR  - https://doi.org/10.2991/978-94-6463-094-7_42
DO  - 10.2991/978-94-6463-094-7_42
ID  - Saberi2022
ER  -