MAESTRO: Masked AutoEncoders for Multimodal, Multitemporal, and Multispectral Earth Observation Data
Not on the map. Why?
Abstract (arXiv)
Self-supervised learning holds great promise for remote sensing, but standard self-supervised methods must be adapted to the unique characteristics of Earth observation data. We take a step in this direction by conducting a comprehensive benchmark of fusion strategies and normalization schemes of reconstruction targets for multimodal, multitemporal, and multispectral Earth observation data. Based on our findings, we introduce MAESTRO, a novel adaptation of the Masked Autoencoder with optimized fusion mechanisms and a normalization scheme that incorporates a spectral prior as a self-supervisory signal. Evaluated on four Earth observation datasets in both intra- and cross-dataset settings, MAESTRO achieves state-of-the-art performance on tasks that strongly rely on multitemporal dynamics, while also remaining competitive on others. Code to reproduce all our experiments is available at https://github.com/ignf/maestro.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 94 of its 118 references that are in the map. First posted on arXiv 14 Aug 2025.
Builds on
94 of 118 references.
References · 94
- 2017Attention Is All You Needcited by 60,361
- 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
- 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
- 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
- 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,403
- 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,968
- 2020Training data-efficient image transformers & distillation through attentioncited by 4,586
- 2025DINOv3cited by 1,414
- 2022VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Trainingcited by 1,322
- 2018Averaging Weights Leads to Wider Optima and Better Generalizationcited by 1,209
- 2021SimMIM: A Simple Framework for Masked Image Modelingcited by 1,153
- 2023Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecturecited by 1,039
38 more
- 2021iBOT: Image BERT Pre-Training with Online Tokenizercited by 903
- 2019Large Batch Optimization for Deep Learning: Training BERT in 76 minutescited by 674
- 2022SatMAE: Pre-training Transformers for Temporal and Multi-Spectral Satellite Imagerycited by 408
- 2022Scale-MAE: A Scale-Aware Masked Autoencoder for Multiscale Geospatial Representation Learningcited by 308
- 2023SpectralGPT: Spectral Remote Sensing Foundation Modelcited by 252
- 2021Seasonal Contrast: Unsupervised Pre-Training from Uncurated Remote Sensing Datacited by 215
- 2023SkySense: A Multi-Modal Remote Sensing Foundation Model Towards Universal Interpretation for Earth Observation Imagerycited by 205
- 2020Geography-Aware Self-Supervised Learningcited by 190
- 2023CROMA: Remote Sensing Representations with Contrastive Radar-Optical Masked Autoencoderscited by 169
- 2022SatlasPretrain: A Large-Scale Dataset for Remote Sensing Image Understandingcited by 167
- 2022Advancing Plain Vision Transformer Towards Remote Sensing Foundation Modelcited by 149
- 2023Foundation Models for Generalist Geospatial Artificial Intelligencecited by 149
- 2024Prithvi-EO-2.0: A Versatile Multi-Temporal Foundation Model for Earth Observation Applicationscited by 144
- 2021Panoptic Segmentation of Satellite Image Time Series with Convolutional Temporal Attention Networkscited by 143
- 2024Scalable Pre-training of Large Autoregressive Image Modelscited by 115
- 2023Towards Geospatial Foundation Models via Continual Pretrainingcited by 109
- 2022On the SDEs and Scaling Rules for Adaptive Gradient Algorithmscited by 104
- 2024Neural Plasticity-Inspired Multimodal Foundation Model for Earth Observationcited by 104
- 2025TerraMind: Large-Scale Generative Multimodality for Earth Observationcited by 104
- 2023Lightweight, Pre-trained Transformers for Remote Sensing Timeseriescited by 103
- 2024Rethinking Transformers Pre-training for Multi-Spectral Satellite Imagerycited by 83
- 2024Mission Critical -- Satellite Data is a Distinct Modality in Machine Learningcited by 77
- 2024MMEarth: Exploring Multi-Modal Pretext Tasks For Geospatial Representation Learningcited by 77
- 2024AnySat: One Earth Observation Model for Many Resolutions, Scales, and Modalitiescited by 74
- 2024HyperSIGMA: Hyperspectral Intelligence Comprehension Foundation Modelcited by 73
- 2024Exponential Moving Average of Weights in Deep Learning: Dynamics and Benefitscited by 71
- 2024MTP: Advancing Remote Sensing Foundation Model via Multi-Task Pretrainingcited by 66
- 2024OmniSat: Self-Supervised Modality Fusion for Earth Observationcited by 63
- 2023A Billion-scale Foundation Model for Remote Sensing Imagescited by 62
- 2023Very high resolution canopy height maps from RGB imagery using self-supervised vision transformer and convolutional decoder trained on Aerial Lidarcited by 60
- 2023ViTs for SITS: Vision Transformers for Satellite Image Time Seriescited by 56
- 2020Lightweight Temporal Self-Attention for Classifying Satellite Image Time Seriescited by 55
- 2025Towards a Unified Copernicus Foundation Model for Earth Visioncited by 53
- 2024Automatic Data Curation for Self-Supervised Learning: A Clustering-Based Approachcited by 50
- 2021Multi-Modal Temporal Attention Models for Crop Mapping from Satellite Time Seriescited by 49
- 2023CMID: A Unified Self-Supervised Learning Framework for Remote Sensing Image Understandingcited by 48
- 2023Feature Guided Masked Autoencoder for Self-supervised Learning in Remote Sensingcited by 48
- 2022MixMAE: Mixed and Masked Autoencoder for Efficient Pretraining of Hierarchical Vision Transformerscited by 47
50 most cited of 94 All 94 (JSON)
Built on by
Citing papers per year: 2026: 2.
2 papers.