TY - GEN
T1 - Leveraging Bioclimatic Context for Supervised and Self-supervised Land Cover Classification
AU - Leonhardt, Johannes
AU - Drees, Lukas
AU - Gall, Jürgen
AU - Roscher, Ribana
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.
PY - 2024
Y1 - 2024
N2 - Modern neural networks achieve state-of-the-art results on land cover classification from satellite imagery, as is the case for almost all vision tasks. One of the main challenges in this context is dealing with geographic variability in both image and label distributions. To tackle this problem, we study the effectiveness of incorporating bioclimatic information into neural network training and prediction. Such auxiliary data can easily be extracted from freely available rasters at satellite images’ georeferenced locations. We compare two methods of incorporation, learned embeddings and conditional batch normalization, to a bioclimate-agnostic baseline ResNet18. In our experiments on the EuroSAT and BigEarthNet datasets, we find that especially the use of conditional batch normalization improves the network’s overall accuracy, generalizability, as well as training efficiency, in both a supervised and a self-supervised learning setup. Code and data are publicly available at https://t.ly/NDQFF.
AB - Modern neural networks achieve state-of-the-art results on land cover classification from satellite imagery, as is the case for almost all vision tasks. One of the main challenges in this context is dealing with geographic variability in both image and label distributions. To tackle this problem, we study the effectiveness of incorporating bioclimatic information into neural network training and prediction. Such auxiliary data can easily be extracted from freely available rasters at satellite images’ georeferenced locations. We compare two methods of incorporation, learned embeddings and conditional batch normalization, to a bioclimate-agnostic baseline ResNet18. In our experiments on the EuroSAT and BigEarthNet datasets, we find that especially the use of conditional batch normalization improves the network’s overall accuracy, generalizability, as well as training efficiency, in both a supervised and a self-supervised learning setup. Code and data are publicly available at https://t.ly/NDQFF.
KW - Conditional Batch Normalization
KW - Data Shift
KW - Land Cover Classification
KW - Multi-Modal Learning
KW - Remote Sensing
UR - https://www.scopus.com/pages/publications/85189542425
U2 - 10.1007/978-3-031-54605-1_15
DO - 10.1007/978-3-031-54605-1_15
M3 - Conference contribution
AN - SCOPUS:85189542425
SN - 9783031546044
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 227
EP - 242
BT - Pattern Recognition - 45th DAGM German Conference, DAGM GCPR 2023, Proceedings
A2 - Köthe, Ullrich
A2 - Rother, Carsten
PB - Springer Science and Business Media Deutschland GmbH
T2 - 45th Annual Conference of the German Association for Pattern Recognition, DAGM-GCPR 2023
Y2 - 19 September 2023 through 22 September 2023
ER -