Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2506.18587 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909656924815360 |
|---|---|
| author | Saget, Antoine Lafabregue, Baptiste Cornuéjols, Antoine Gançarski, Pierre |
| author_facet | Saget, Antoine Lafabregue, Baptiste Cornuéjols, Antoine Gançarski, Pierre |
| contents | Given the abundance of unlabeled Satellite Image Time Series (SITS) and the scarcity of labeled data, contrastive self-supervised pretraining emerges as a natural tool to leverage this vast quantity of unlabeled data. However, designing effective data augmentations for contrastive learning remains challenging for time series. We introduce a novel resampling-based augmentation strategy that generates positive pairs by upsampling time series and extracting disjoint subsequences while preserving temporal coverage. We validate our approach on multiple agricultural classification benchmarks using Sentinel-2 imagery, showing that it outperforms common alternatives such as jittering, resizing, and masking. Further, we achieve state-of-the-art performance on the S2-Agri100 dataset without employing spatial information or temporal encodings, surpassing more complex masked-based SSL frameworks. Our method offers a simple, yet effective, contrastive learning augmentation for remote sensing time series. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_18587 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Resampling Augmentation for Time Series Contrastive Learning: Application to Remote Sensing Saget, Antoine Lafabregue, Baptiste Cornuéjols, Antoine Gançarski, Pierre Computer Vision and Pattern Recognition Given the abundance of unlabeled Satellite Image Time Series (SITS) and the scarcity of labeled data, contrastive self-supervised pretraining emerges as a natural tool to leverage this vast quantity of unlabeled data. However, designing effective data augmentations for contrastive learning remains challenging for time series. We introduce a novel resampling-based augmentation strategy that generates positive pairs by upsampling time series and extracting disjoint subsequences while preserving temporal coverage. We validate our approach on multiple agricultural classification benchmarks using Sentinel-2 imagery, showing that it outperforms common alternatives such as jittering, resizing, and masking. Further, we achieve state-of-the-art performance on the S2-Agri100 dataset without employing spatial information or temporal encodings, surpassing more complex masked-based SSL frameworks. Our method offers a simple, yet effective, contrastive learning augmentation for remote sensing time series. |
| title | Resampling Augmentation for Time Series Contrastive Learning: Application to Remote Sensing |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2506.18587 |