Holzmann M, Davari A, Seehaus T, Braun M, Maier A, Christlein V (2021)
Publication Type: Conference contribution, Original article
Publication year: 2021
Original Authors: Michael Holzmann, Amirabbas Davari, Thorsten Seehaus, Matthias Braun, Andreas Maier, Vincent Christlein
URI: https://arxiv.org/abs/2101.03247
DOI: 10.1109/IGARSS47720.2021.9555067
An essential climate variable to determine the tidewater glacier status is the location of the calving front position and the separation of seasonal variability from long-term trends. Previous studies have proposed deep learning-based methods to semi-automatically delineate the calving fronts of tidewater glaciers. They used U-Net to segment the ice and non-ice regions and extracted the calving fronts in a post-processing step. In this work, we show a method to segment the glacier calving fronts from SAR images in an end-to-end fashion using Attention U-Net. The main objective is to investigate the attention mechanism in this application. Adding attention modules to the state-of-the-art U - N et network lets us analyze the learning process by extracting its attention maps. We use these maps as a tool to search for proper hyperparameters and loss functions in order to generate higher qualitative results. Our proposed attention U-Net performs comparably to the standard U-Net while providing additional insight into those regions on which the network learned to focus more. In the best case, the attention U-Net achieves a 1.5 % better Dice score compared to the canonical U-Net with a glacier front line prediction certainty of up to 237.12 meters.
APA:
Holzmann, M., Davari, A., Seehaus, T., Braun, M., Maier, A., & Christlein, V. (2021). Glacier Calving Front Segmentation Using Attention U-Net. In Proceedings of the 2021 IEEE International Symposium on Geoscience and Remote Sensing (IGARSS). Brussels, BE.
MLA:
Holzmann, Michael, et al. "Glacier Calving Front Segmentation Using Attention U-Net." Proceedings of the 2021 IEEE International Symposium on Geoscience and Remote Sensing (IGARSS), Brussels 2021.
BibTeX: Download