Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Lightweight, Pre-trained Transformers for Remote Sensing Timeseries

About

Machine learning methods for satellite data have a range of societally relevant applications, but labels used to train models can be difficult or impossible to acquire. Self-supervision is a natural solution in settings with limited labeled data, but current self-supervised models for satellite data fail to take advantage of the characteristics of that data, including the temporal dimension (which is critical for many applications, such as monitoring crop growth) and availability of data from many complementary sensors (which can significantly improve a model's predictive performance). We present Presto (the Pretrained Remote Sensing Transformer), a model pre-trained on remote sensing pixel-timeseries data. By designing Presto specifically for remote sensing data, we can create a significantly smaller but performant model. Presto excels at a wide variety of globally distributed remote sensing tasks and performs competitively with much larger models while requiring far less compute. Presto can be used for transfer learning or as a feature extractor for simple models, enabling efficient deployment at scale.

Gabriel Tseng, Ruben Cartuyvels, Ivan Zvonkov, Mirali Purohit, David Rolnick, Hannah Kerner• 2023

Related benchmarks

TaskDatasetResultRank
ClassificationTreeSatAI-TS
F1 Score67.81
75
Semantic segmentationAustrian Crop (30% labels)
mIoU27.93
25
Semantic segmentationAustrian Crop All labels
mIoU34.04
25
RegressionBorneo CHM (1% labels)
RMSE36.43
25
RegressionBorneo CHM (30% labels)
RMSE20.95
25
Semantic segmentationAustrian Crop (1% labels)
mIoU14.25
25
RegressionBorneo CHM (All labels)
RMSE17.88
25
Crop SegmentationPASTIS
mIoU28.1
12
Crop type classificationCropGlobe FRA → BEL
Overall Accuracy96.35
7
Crop type classificationCropGlobe USA → FRA
Overall Accuracy62.24
7
Showing 10 of 27 rows

Other info

Code

Follow for update