Pith. sign in

REVIEW 4 major objections 5 minor 10 cited by

AnySat: One Earth Observation Model for Many Resolutions, Scales, and Modalities

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A single self-supervised model pretrained on eleven heterogeneous sensors — 0.2 m to 250 m resolution — beats task-specific specialists on nine downstream Earth-observation benchmarks.

desk verdict Solid architectural contribution, but the SOTA claims run ahead of the evidence; the protocol asymmetries are disclosed but they still undercut the headline numbers. read the letter →

arxiv 2412.14123 v3 pith:L2G3AG4H submitted 2024-12-18 cs.CV

Guillaume Astruc , Nicolas Gonthier , Clement Mallet , Loic Landrieu This is my paper · ORCID
classification cs.CV
keywords Earthobservationfoundationmodeljointembeddingpredictivearchitectureself-supervisedlearningmultimodalremotesensingmulti-resolutionscale-adaptiveencoderGeoPlex
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Earth observation data arrive in wildly different forms — 20 cm aerial photos, 10 m radar and optical time series, 250 m climate bands — and until now each dataset or task has needed its own model with a fixed input format. AnySat is a single architecture pretrained in a self-supervised way on all of these at once: five datasets, eleven sensors, resolutions from 0.2 to 250 m, without resizing or per-dataset retraining. After pretraining, the same model fine-tunes or linearly probes to state-of-the-art results on nine monitoring tasks, including six external datasets whose sensor combinations never appeared in pretraining. If the claim holds, one Earth-observation foundation model can replace the current zoo of per-dataset models and adapt to a new sensor with little more than a new input projector.

What carries the argument

The load-bearing mechanism is the scale-adaptive patch encoder coupled with a multimodal JEPA training objective. A tile is cut into patches of a fixed physical size $P \times P$ meters, shared by all modalities; each patch is split into fixed-size sub-patches of $\delta_m$ pixels, embedded by a modality-specific projector (with a lightweight temporal attention module for time series), and aggregated by a shared spatial transformer whose token count — but never its embedding size — changes with resolution and patch size. This turns 'many resolutions and modalities' into a token-count problem solvable with a single weight set. The training signal is then the semantics shared across sensors: a contrastive loss aligns same-patch embeddings from different modalities, and a JEPA prediction loss forces a masked student to reproduce the unmasked teacher's embeddings for dropped patches.

What would settle it

A decisive experiment: pretrain AnySat on GeoPlex with the European and US datasets removed (or on an equal-size corpus drawn only from the tropics and southern hemisphere), then fine-tune on the same external benchmarks — SICKLE (India), BraDD-S1TS (Amazon), Sen1Floods11 (global). If the fine-tuned gains over task-specific baselines largely survive, the representations are genuinely resolution- and modality-driven; if they collapse, the reported external gains owe more to geographic overlap or dataset-specific shortcuts. A cheaper second check: measure linear-probe performance on a geographically held-out continent that shares no biome with GeoPlex.

Watch

Extended reading notes

Core claim

AnySat's central claim is that spatial alignment alone can carry multimodal Earth-observation learning: when different sensors observe the same ground patch, their representations should agree and should be predictable from one another in feature space, with no need to reconstruct pixels. The paper realizes this with a multimodal Joint Embedding Predictive Architecture (JEPA): a student network processes heavily masked and modality-dropped patches and must reproduce the embeddings produced by an unmasked teacher whose weights are an exponential moving average of the student's, trained by an L2 prediction loss plus a contrastive loss that pulls same-patch, different-modality embeddings together. Because prediction happens in latent space, no modality-specific decoders are required, and the scale-adaptive patch encoder means one shared weight set handles resolutions from 0.2 m to 250 m. Pretrained self-supervised on GeoPlex (five datasets, eleven sensors, 249K km²), AnySat is then fine-tuned or linearly probed to state-of-the-art results on nine downstream tasks — land cover, crop type, tree species, flood, burn scar, deforestation, and climate zone mapping — including six external datasets whose sensor configurations never appeared during pretraining.

Load-bearing premise

The load-bearing premise is that GeoPlex — 249K km² concentrated in Europe and the United States — is representative enough of global land cover and imaging conditions that representations trained on it transfer to India, the Amazon, and worldwide flood scenes, rather than the external gains being geographic or dataset-specific shortcuts.

Editorial extensions

If this is right

  • Adaptation cost collapses: a linear probe with about 6,000 trainable parameters beats or matches specialized models and a 630M-parameter foundation model on flood segmentation and Amazon deforestation detection, so new monitoring tasks become cheap to deploy.
  • Unseen sensors and configurations work after minimal surgery: single-date Sentinel pairs, single-pixel time series, and a wholly unseen sensor (HLS) all reach state-of-the-art results, the last via a newly initialized projector.
  • Pretraining pays off most where labels are scarce — the largest gains are on the smallest labeled datasets (TreeSatAI-TS, PASTIS classification) — so label-poor environmental monitoring stands to benefit most.
  • GeoPlex is extensible: because each training step samples a random dataset and patch size, new datasets can be folded into the same pretraining run, letting every added region or sensor sharpen the shared backbone.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper leaves its own geographic stress test implicit: GeoPlex skews heavily toward Europe and the United States, so the decisive experiment is whether the external-benchmark wins in India, the Amazon, and global floods survive when pretraining is rebuilt from tropical and southern-hemisphere data alone; that test would separate genuine resolution/modality generalization from geography-driven s
  • The recipe — align modalities through shared ground location, predict in embedding space rather than pixel space — is not obviously satellite-specific; applying the identical architecture to multimodal medical imaging, where CT, MRI, and ultrasound view the same anatomy, would be a direct transfer test the authors do not run.
  • The ablation in which removing the contrastive loss costs 4.3 F1 points on classification while removing masking costs little suggests a leaner recipe — strong cross-modal contrast with minimal masking — might reach similar representations at lower pretraining cost; this follows from the paper's own numbers but is not an experiment it performs.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. AnySat proposes a single JEPA-based multimodal Earth observation model whose scale-adaptive patch encoder accepts heterogeneous resolutions, patch sizes, and modalities. The model is pretrained self-supervised on GeoPlex, a newly compiled collection of five datasets and eleven sensors, and then fine-tuned or linearly probed on nine downstream tasks, including six external datasets. The paper reports state-of-the-art or near-state-of-the-art results across these tasks and provides ablations of patch dropping, the contrastive loss, and subpatch-based segmentation.

Significance. If the evaluation were strictly apples-to-apples, the result would be a meaningful step: one architecture simultaneously trained on highly heterogeneous EO data and transferred to unseen sensor configurations without per-dataset retraining. The paper is unusually transparent about several limitations, including the use of the SICKLE validation set, the inclusion of all PASTIS-HD tiles in self-supervised pretraining, and the modification of competitor input layers. The code and models are released, and the ablation study provides useful evidence for the main design choices. The architecture itself is simple and GeoPlex is a valuable resource. However, the protocol asymmetries listed below currently prevent the central 'state-of-the-art' claim from being fully established; I do not consider geographic coverage of GeoPlex to be the most fragile point, since the external transfer results themselves are evidence of generalization.

major comments (4)
  1. [Section 4.1 and Appendix D (SICKLE)] The paper states that SICKLE's test set has not been released and that the authors use the validation set, while the baseline numbers in Table B are taken from earlier papers. If those baselines were reported on the official test split, the claimed +3.6 mIoU improvement is not an apples-to-apples comparison. Please report AnySat on the published SICKLE test split if it is now available, or rerun the comparison methods on the same validation split and report all numbers under an identical protocol.
  2. [Section 4.1 and 4.2 (PASTIS-HD)] PASTIS-HD is included in GeoPlex 'in its entirety (without labels),' and PASTIS-HD is normally evaluated by 5-fold cross-validation. The SSL pretraining therefore sees the imagery of the test folds, and the reported +2.8 macro-F1 / +0.2 mIoU gains may be transductive rather than evidence of generalizable representation learning. The ablation section describes pretraining on folds 1 through 4 and evaluating on a held-out fold; this protocol should be used for the main PASTIS-HD results as well, and both versions should be compared explicitly.
  3. [Section 4.1 (Competing Methods)] For foundation models that cannot consume the target modalities, the paper modifies their input layers to match the target number of spectral bands. This is a known source of disadvantage for pretrained models, especially when the pretraining was tied to native sensors, such as CROMA with Sentinel-2 and ScaleMAE/SatMAE with RGB input. The paper should justify each modification, report the exact adaptation procedure, or rely on published baselines with their original protocols; otherwise the comparisons in Tables A and B are not controlled.
  4. [Sections 4.2 and Appendix A (Tables A-B)] The reported margins include several very small differences, notably PASTIS-HD segmentation (+0.2 mIoU), FLAIR (within roughly 0.5-1.8 points of the best baseline), and So2Sat (59.1 vs. DOFA's 59.3). The paper gives no error bars, confidence intervals, or multi-seed runs, so these differences are within plausible run-to-run noise. Quantitative state-of-the-art claims need variance estimates or at least a clear statement that such small differences are not being used as evidence.
minor comments (5)
  1. [Section 4.2 and Table 2] The text says the ablation is performed on Fold 5 of PASTIS-HD, while the Table 2 caption says Fold 1; please reconcile this inconsistency.
  2. [Introduction and Appendix C] The introduction claims that 'over 75% of the learnable parameters' are shared, while Appendix C reports 73.6% modality-agnostic parameters; the numbers should be consistent.
  3. [Appendix C (Optimization Parameters)] The sentence 'We set he contrastive temperatureγ to 0.1 to n Eq. X' contains a typo and should read 'We set the contrastive temperature gamma to 0.1 in Eq. (2)'.
  4. [Table 1] The column header layout in Table 1 is very hard to follow; please reformat it so that each benchmark and metric is clearly aligned with its column.
  5. [Section 4.1 and Table C (PLANTED)] The text notes that only 1.3 of the 2.3 million PLANTED images used in the original paper are publicly available; please clarify explicitly whether GeoPlex uses only the public 1.3 million images for pretraining.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: AnySat's external benchmarks are held out; OmniSat self-citations are architectural and non-reductive.

full rationale

The derivation chain is not circular. AnySat's central claim is that a single model pretrained on GeoPlex can be fine-tuned or linearly probed to reach state-of-the-art results on 9 downstream tasks, including 6 external datasets. Those external datasets (BraDD-S1TS, TimeSen2Crop, Sen1Floods11, So2Sat, HLS Burn Scar, SICKLE) are not used for SSL pretraining, and the reported metrics are compared against published baselines rather than derived from the pretraining objective. The method does cite the authors' own OmniSat for the combiner architecture and contrastive loss: 'We use the cross-attention-based architecture proposed by OmniSat [7, 3.1]' and 'a contrastive loss inspired by OmniSat [7]'. This is self-citation, but it is not load-bearing in a circular sense: OmniSat is a published ECCV paper, and the paper explicitly compares AnySat against OmniSat as a baseline and reports higher scores, so the cited component is independently testable rather than an assumed conclusion. No equation in the paper reduces a predicted quantity to a fitted parameter by construction; the JEPA loss, contrastive loss, and downstream probing are all standard training/evaluation loops. The protocol limitations noted in the paper—PASTIS-HD is included 'in its entirety (without labels)' in GeoPlex, SICKLE's test set is unreleased so the validation set is used, and some baselines had modified input layers—are real evaluation-asymmetry concerns that weaken the SOTA comparisons, but they do not constitute circularity: the model does not fit the labels of those splits by construction. Therefore the paper is largely self-contained in its empirical claims, with only minor self-citation that does not drive the conclusion.

Assumptions & free parameters 6 free parameters · 6 assumptions · 0 invented entities

The central claim rests on standard self-supervised learning assumptions and on hand-set hyperparameters rather than on derived constants from first principles. No new physical entities are introduced; the only new objects are learned network parameters such as the mask and drop tokens, which are part of the architecture rather than independent postulates.

free parameters (6)
  • contrastive temperature tau = 0.1
    Set by hand in Appendix C; controls the sharpness of the contrastive loss in Eq. 2.
  • EMA decay = 0.996
    Set in Appendix C; controls the teacher network update rate in the JEPA training loop.
  • learning rate = 5e-5
    Used for pretraining and all fine-tuning experiments; set in Appendix C.
  • sub-patch size delta_m = 1 pixel for VHR images, 10 pixels for time series
    Chosen per modality in Section 4.1; determines the token granularity of the scale-adaptive encoder.
  • per-dataset patch size sets P_d = e.g., TreeSatAI {10,20,30} m, PASTIS {40,80,160} m, FLAIR {10,20,50} m
    Selected per dataset in Tab. C; controls spatial context and number of sub-patch tokens.
  • masking schedule = 5 dropped rectangular regions, 50% temporal masking
    Hand-set in Section 3.2; defines the student corruption process for JEPA pretraining.
assumptions (6)
  • domain assumption JEPA feature-predictive pretraining yields semantically useful representations for Earth observation data.
    Adopted from Assran et al. and applied to EO in Section 3.2; the downstream results are the test of this assumption.
  • domain assumption Spatially aligned observations of the same patch from different sensors share underlying semantics.
    Stated in Sections 1 and 3.2 and encoded in the contrastive loss; false when acquisitions differ in time, weather, or season.
  • domain assumption The OmniSat cross-attention combiner is an effective fusion module when embedded in a JEPA framework.
    Section 3.1 reuses the combiner from the authors' prior OmniSat paper without independent verification in this new context.
  • domain assumption Scale-adaptive positional encodings based on ground sampling distance generalize across resolutions and patch sizes.
    Appendix C builds on Scale-MAE; the paper does not prove this property, only tests it empirically.
  • domain assumption GeoPlex is representative enough of global Earth observation diversity to enable transfer to unseen regions and sensors.
    Section 4.1 and Fig. 4 show a dataset concentrated in Europe and the US; external global datasets are the evidence for this assumption.
  • domain assumption The EMA teacher provides stable and useful prediction targets.
    Inherited from the JEPA and MoCo literature and used in Section 3.2; no analysis of target quality is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AnySat: One Earth Observation Model for Many Resolutions, Scales, and Modalities." pith.science (2026). https://pith.science/paper/L2G3AG4H

@misc{pith2026241214123,
  author       = {Pith},
  title        = {Pith review of: AnySat: One Earth Observation Model for Many Resolutions, Scales, and Modalities},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/L2G3AG4H}},
  note         = {Machine review of arXiv:2412.14123}
}
abstract

Geospatial models must adapt to the diversity of Earth observation data in terms of resolutions, scales, and modalities. However, existing approaches expect fixed input configurations, which limits their practical applicability. We propose AnySat, a multimodal model based on joint embedding predictive architecture (JEPA) and scale-adaptive spatial encoders, allowing us to train a single model on highly heterogeneous data in a self-supervised manner. To demonstrate the advantages of this unified approach, we compile GeoPlex, a collection of 5 multimodal datasets with varying characteristics and $11$ distinct sensors. We then train a single powerful model on these diverse datasets simultaneously. Once fine-tuned or probed, we reach state-of-the-art results on the test sets of GeoPlex and for 6 external datasets across various environment monitoring tasks: land cover mapping, tree species identification, crop type classification, change detection, climate type classification, and segmentation of flood, burn scar, and deforestation. The code and models are available at https://github.com/gastruc/AnySat.

Figures

Figures reproduced from arXiv: 2412.14123 by the authors.

Figure 1
Figure 1. Multi-Dataset Training. For the first time, a single model can be pretrained simultaneously on a collection of Earth Observation datasets with heterogeneous resolutions, scales, and modalities. The resulting model can be fine-tuned to achieve state￾of-the-art results for a wide variety of data types and tasks. the purpose of foundation models. EO foundation mod￾els should be able to seamlessly integrate new datasets… view at source ↗
Figure 2
Figure 2. Scale-Adaptive Patch Encoding. We consider a patch x m p of resolution ∆m = P/Rm pixels. We first split x m p into sub-patches of size δm pixels, which are mapped by a modality-specific projector ϕ proj m to a E-dimensional embedding. Then, a shared spatial transformer module ϕ trans combines all sub-patches into a vector of size E. As the sub-patch size δm is fixed, the patch sizes ∆m only influences the number of … view at source ↗
Figure 3
Figure 3. Architecture of AnySat. We begin each iteration by randomly selecting a dataset among GeoPlex and sampling a tile. Each available modality is divided into spatially aligned patches of size P. The student network’s patch encoder ϕ patch S embeds each patch and we apply a contrastive loss to encourage spatial consistency across modalities. We then apply dropping and masking : some patches have all modalities removed (… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Datasets Considered. GeoPlex is composed of 5 diverse dataset spanning the entire world, with a higher concentration in Europe and the US where open-data are more abundant. We also consider external evaluation datasets with a more diverse spread. 2. Randomly select a p…
Figure 5
Figure 5. Figure 5: Quantitative Evaluation. We evaluate AnySat across 9 open-access datasets and for four tasks: multilabel classification (classif), semantic segmentation (semseg), pixel-wise change detection (chgdet), and pixel-wise regression (regression). For clarity, we only visuali…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 10 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. TESSERA v2: Scaling Pixel-wise Earth Foundation Models

    cs.CV 2026-07 conditional novelty 7.0 of 10

    Downstream-driven scaling of pixel-wise Barlow Twins EO models favors large encoders and matched data over projectors, and distillation yields compact Matryoshka students that lead multi-task embedding benchmarks.

  2. Galileo: Learning Global & Local Features of Many Remote Sensing Modalities

    cs.CV 2025-02 conditional novelty 7.0 of 10

    A single multimodal transformer, Galileo, jointly learns global and local features from optical, radar, elevation, weather, and land-cover inputs and outperforms specialized models on eleven benchmarks.

  3. CanadaFireSat: Toward high-resolution wildfire forecasting with multiple modalities

    cs.CV 2025-06 conditional novelty 6.5 of 10

    Introduces a multi-modal 100m wildfire forecasting benchmark for Canada and shows deep learning models benefit from fusing Sentinel-2 imagery with environmental predictors.

  4. Be the Change You Want to See: Revisiting Remote Sensing Change Detection Practices

    cs.CV 2025-07 conditional novelty 6.0 of 10

    A simple, well-optimized model (BTC) beats or matches state-of-the-art change detection methods on six datasets, indicating that fundamental design choices matter more than architectural novelty.

  5. TerraFM: A Scalable Foundation Model for Unified Multisensor Earth Observation

    cs.CV 2025-06 conditional novelty 6.0 of 10

    A multimodal satellite foundation model trained with contrastive learning, modality-aware patch embeddings, cross-attention fusion, and a dual-centering regularizer achieves state-of-the-art results on GEO-Bench and C...

  6. SkySense V2: A Unified Foundation Model for Multi-modal Remote Sensing

    cs.CV 2025-07 conditional novelty 5.0 of 10

    A unified multi-modal remote sensing foundation model with adaptive patch merging, modality prompt tokens, mixture of experts, and query-based semantic aggregation contrastive learning outperforms SkySense by 1.8 poin...

  7. Deploying Geospatial Foundation Models in the Real World: Lessons from WorldCereal

    cs.LG 2025-07 conditional novelty 5.0 of 10

    A structured protocol for deploying geospatial foundation models is introduced and validated in WorldCereal, where fine-tuned Presto outperforms a fully-supervised CatBoost baseline in crop mapping.

  8. High-Resolution Live Fuel Moisture Content (LFMC) Maps for Wildfire Risk from Multimodal Earth Observation Data

    cs.LG 2025-06 conditional novelty 5.0 of 10

    Fine-tuning the pretrained Galileo model on the Globe-LFMC dataset yields 10 m wall-to-wall live fuel moisture maps with RMSE 18.91, about 20% better than a randomly initialized model.

  9. Large scale cross-regional remote sensing flood monitoring framework for operative mapping and impact analysis

    cs.CV 2026-07 conditional novelty 4.0 of 10

    Under limited Russian flood labels, multimodal U-Net++ (F1 0.84) outperforms fine-tuned AnySat for water mapping, and the masks feed an EMERCOM-style damage pipeline that matches Tulun 2019 official area and exposure ...

  10. Scalable and Trustworthy Earth Observation Foundation Models

    cs.LG 2026-07 conditional novelty 3.0 of 10

    Remote-sensing foundation models need domain-specific design and evaluation around measurement physics and decision constraints; benchmark accuracy alone is insufficient for trustworthy EO deployment.

Reference graph

Works this paper leans on

83 extracted references · 68 canonical work pages · cited by 10 Pith papers

  1. [1]

    https : / / lightning - flash

    Lightning: LinearWarmupCosineAnneal- ingLR. https : / / lightning - flash . readthedocs . io / en / stable / api / generated / flash . core . optimizers . LinearWarmupCosineAnnealingLR . html . Accessed: 2024-11-20. 16

    2024

  2. [2]

    org / docs / stable / generated / torch

    PyTorch: ReduceLROnPlateau. org / docs / stable / generated / torch . optim . lr _ scheduler . ReduceLROnPlateau . html # torch . optim . lr _ scheduler . ReduceLROnPlateau. Accessed: 2024-02-

    2024

  3. [3]

    TreeSatAI Benchmark Archive: A multi-sensor, multi-label dataset for tree species classi- fication in remote sensing

    Steve Ahlswede, Christian Schulz, Christiano Gava, Patrick Helber, Benjamin Bischke, Michael F ¨orster, Florencia Arias, J ¨orn Hees, Beg ¨um Demir, and Bir- git Kleinschmit. TreeSatAI Benchmark Archive: A multi-sensor, multi-label dataset for tree species classi- fication in remote sensing. Earth System Science Data Discussions, 2022. 6, 8, 17, 18

    2022

  4. [4]

    AI2-S2-NAIP

    allenai.org. AI2-S2-NAIP. https://huggingface.co/datasets/allenai/s2-naip,

    work page

  5. [5]

    ViViT: A video vision transformer

    Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Luˇci´c, and Cordelia Schmid. ViViT: A video vision transformer. In CVPR, 2021. 7, 15

    work page 2021

  6. [6]

    Self-supervised learning from images with a joint-embedding predictive architecture

    Mahmoud Assran, Quentin Duval, Ishan Misra, Piotr Bojanowski, Pascal Vincent, Michael Rabbat, Yann Le- Cun, and Nicolas Ballas. Self-supervised learning from images with a joint-embedding predictive architecture. In CVPR, 2023. 2, 4, 8

    work page 2023

  7. [7]

    Omnisat: Self-supervised modality fusion for earth observation

    Guillaume Astruc, Nicolas Gonthier, Clement Mallet, and Loic Landrieu. Omnisat: Self-supervised modality fusion for earth observation. In ECCV, 2024. 1, 2, 3, 4, 6, 9, 14, 15, 17, 18

    work page 2024

  8. [8]

    Foundational models defining a new era in vision: A survey and outlook

    Muhammad Awais, Muzammal Naseer, Salman Khan, Rao Muhammad Anwer, Hisham Cholakkal, Mubarak Shah, Ming-Hsuan Yang, and Fahad Shahbaz Khan. Foundational models defining a new era in vision: A survey and outlook. arXiv preprint arXiv:2307.13721,

    arXiv

Show all 83 references
  1. [9]

    Geography-aware self-supervised learning

    Kumar Ayush, Burak Uzkent, Chenlin Meng, Kumar Tanmay, Marshall Burke, David Lobell, and Stefano Ermon. Geography-aware self-supervised learning. In ICCV, 2021. 2

    2021

  2. [10]

    Data2vec: A gen- eral framework for self-supervised learning in speech, vision and language

    Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, and Michael Auli. Data2vec: A gen- eral framework for self-supervised learning in speech, vision and language. In ICML, 2022. 2, 15

    2022

  3. [11]

    SatlasPretrain: A large-scale dataset for remote sensing image under- standing

    Favyen Bastani, Piper Wolters, Ritwik Gupta, Joe Fer- dinando, and Aniruddha Kembhavi. SatlasPretrain: A large-scale dataset for remote sensing image under- standing. In ICCV, 2023. 1, 2, 6, 16

    2023

  4. [12]

    Sen1Floods11: A georeferenced dataset to train and test deep learning flood algorithms for Sentinel-1

    Derrick Bonafilia, Beth Tellman, Tyler Anderson, and Erica Issenberg. Sen1Floods11: A georeferenced dataset to train and test deep learning flood algorithms for Sentinel-1. In CVPR Workshop EarthVision, 2020. 7, 14, 16, 17, 18

    2020

  5. [13]

    Learning representations of satel- lite images from metadata supervision

    Jules Bourcier, Gohar Dashyan, Karteek Alahari, and Jocelyn Chanussot. Learning representations of satel- lite images from metadata supervision. In ECCV, 2024. 2

    2024

  6. [14]

    Classi- fication of imbalanced remote-sensing data by neural networks

    Lorenzo Bruzzone and Sebastiano B Serpico. Classi- fication of imbalanced remote-sensing data by neural networks. Pattern recognition letters, 1997. 16 9

    1997

  7. [15]

    Emerging properties in self-supervised vision transformers

    Mathilde Caron, Hugo Touvron, Ishan Misra, Herv ´e J´egou, Julien Mairal, Piotr Bojanowski, and Armand Joulin. Emerging properties in self-supervised vision transformers. In ICCV, 2021. 2

    2021

  8. [16]

    A simple framework for contrastive learning of visual representations

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. A simple framework for contrastive learning of visual representations. In ICML, 2020. 2

    2020

  9. [17]

    Functional map of the world

    Gordon Christie, Neil Fendley, James Wilson, and Ryan Mukherjee. Functional map of the world. In CVPR, 2018. 7

    2018

  10. [18]

    SatMAE: Pre-training transformers for temporal and multi-spectral satellite imagery

    Yezhen Cong, Samar Khanna, Chenlin Meng, Patrick Liu, Erik Rozi, Yutong He, Marshall Burke, David Lobell, and Stefano Ermon. SatMAE: Pre-training transformers for temporal and multi-spectral satellite imagery. In NeurIPS, 2022. 2, 8, 15, 16

    2022

  11. [19]

    MSResNet: Multiscale residual network via self-supervised learning for water- body detection in remote sensing imagery

    Bo Dang and Yansheng Li. MSResNet: Multiscale residual network via self-supervised learning for water- body detection in remote sensing imagery. Remote Sensing, 2021. 16

    2021

  12. [20]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. An image is worth 16x16 words: Transformers for image recognition at scale. In ICLR, 2020. 3

    2020

  13. [21]

    Paving the way toward foundation models for irregular and unaligned satellite image time series

    Iris Dumeur, Silvia Valero, and Jordi Inglada. Paving the way toward foundation models for irregular and unaligned satellite image time series. arXiv preprint arXiv:2407.08448, 2024. 2

    2024 arXiv

  14. [22]

    Self- supervised spatio-temporal representation learning of satellite image time series

    Iris Dumeur, Silvia Valero, and Jordi Inglada. Self- supervised spatio-temporal representation learning of satellite image time series. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sens- ing, 2024. 2

    2024

  15. [23]

    CROMA: Remote sensing representations with con- trastive radar-optical masked autoencoders

    Anthony Fuller, Koreen Millard, and James R Green. CROMA: Remote sensing representations with con- trastive radar-optical masked autoencoders. InNeurIPS,

  16. [24]

    Datacomp: In search of the next generation of multimodal datasets

    Samir Yitzhak Gadre, Gabriel Ilharco, Alex Fang, Jonathan Hayase, Georgios Smyrnis, Thao Nguyen, Ryan Marten, Mitchell Wortsman, Dhruba Ghosh, Jieyu Zhang, et al. Datacomp: In search of the next generation of multimodal datasets. In NeurIPS, 2024. 1

    2024

  17. [25]

    FLAIR: A country-scale land cover semantic segmentation dataset from multi-source optical imagery

    Anatol Garioud, Nicolas Gonthier, Loic Landrieu, Apolline De Wit, Marion Valette, Marc Poup ´ee, S´ebastien Giordano, and Boris Wattrelos. FLAIR: A country-scale land cover semantic segmentation dataset from multi-source optical imagery. In NeurIPS Dataset and Benchmark, 2023....

    2023

  18. [26]

    Lightweight temporal self-attention for classifying satellite images time series

    Vivien Sainte Fare Garnot and Loic Landrieu. Lightweight temporal self-attention for classifying satellite images time series. In Advanced Analytics and Learning on Temporal Data: ECML PKDD Work- shop, 2020. 3, 15, 16

    2020

  19. [27]

    Panoptic segmentation of satellite image time series with convo- lutional temporal attention networks

    Vivien Sainte Fare Garnot and Loic Landrieu. Panoptic segmentation of satellite image time series with convo- lutional temporal attention networks. In ICCV, 2021. 6, 14, 15, 16, 17

    2021

  20. [28]

    Multi-modal temporal attention models for crop mapping from satellite time series

    Vivien Sainte Fare Garnot, Loic Landrieu, and Nesrine Chehata. Multi-modal temporal attention models for crop mapping from satellite time series. ISPRS Journal of Photogrammetry and Remote Sensing, 2022. 15, 18

    2022

  21. [29]

    Unsupervised representation learning by predicting im- age rotations

    Spyros Gidaris, Praveer Singh, and Nikos Komodakis. Unsupervised representation learning by predicting im- age rotations. In ICLR, 2018. 2

    2018

  22. [30]

    Bootstrap your own latent-a new approach to self-supervised learning

    Jean-Bastien Grill, Florian Strub, Florent Altch ´e, Corentin Tallec, Pierre Richemond, Elena Buchatskaya, Carl Doersch, Bernardo Avila Pires, Zhaohan Guo, Mo- hammad Gheshlaghi Azar, et al. Bootstrap your own latent-a new approach to self-supervised learning. In NeurIPS, 2020. 2

    2020

  23. [31]

    Skysense: A multi-modal remote sensing foundation model towards universal interpretation for earth observation imagery

    Xin Guo, Jiangwei Lao, Bo Dang, Yingying Zhang, Lei Yu, Lixiang Ru, Liheng Zhong, Ziyuan Huang, Kang Wu, Dingxiang Hu, et al. Skysense: A multi-modal remote sensing foundation model towards universal interpretation for earth observation imagery. In CVPR,

  24. [32]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In CVPR, 2016. 15

    2016

  25. [33]

    Momentum contrast for unsupervised visual representation learning

    Kaiming He, Haoqi Fan, Yuxin Wu, Saining Xie, and Ross Girshick. Momentum contrast for unsupervised visual representation learning. In CVPR, 2020. 2, 4, 15

    2020

  26. [34]

    Masked autoencoders are scalable vision learners

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Doll´ar, and Ross Girshick. Masked autoencoders are scalable vision learners. In CVPR, 2022. 2

    2022

  27. [35]

    Spec- tralGPT: Spectral remote sensing foundation model

    Danfeng Hong, Bing Zhang, Xuyang Li, Yuxuan Li, Chenyu Li, Jing Yao, Pedram Ghamisi, Naoto Yokoya, Hao Li, Xiuping Jia, Antonio Plaza, Paolo Gamba, Jon Atli Benediktsson, and Jocelyn Chanussot. Spec- tralGPT: Spectral remote sensing foundation model. TPAMI, 2024. 15

    2024

  28. [36]

    Geospa- tial foundation models for image analysis: Evaluating and enhancing NASA-IBM Prithvi’s domain adaptabil- ity

    Chia-Yu Hsu, Wenwen Li, and Sizhe Wang. Geospa- tial foundation models for image analysis: Evaluating and enhancing NASA-IBM Prithvi’s domain adaptabil- ity. International Journal of Geographical Information Science, 2024. 2

    2024

  29. [37]

    USat: A unified self-supervised encoder for multi-sensor satellite imagery

    Jeremy Irvin, Lucas Tao, Joanne Zhou, Yuntao Ma, Langston Nashold, Benjamin Liu, and Andrew Y Ng. USat: A unified self-supervised encoder for multi-sensor satellite imagery. arXiv preprint arXiv:2312.02199, 2023. 8

    2023 arXiv

  30. [38]

    Foundation models for general- ist geospatial artificial intelligence

    Johannes Jakubik, S Roy, CE Phillips, P Fraccaro, D Godwin, B Zadrozny, D Szwarcman, C Gomes, G Nyir- 10 jesy, B Edwards, et al. Foundation models for general- ist geospatial artificial intelligence. URL https://arxiv. org/abs/2310.18660. 1, 2, 8, 15, 16

    arXiv

  31. [39]

    Deforestation detection in the Amazon with Sentinel-1 SAR image time series

    Kaan Karaman, V Sainte Fare Garnot, and Jan Dirk Wegner. Deforestation detection in the Amazon with Sentinel-1 SAR image time series. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial In- formation Sciences, 2023. 7, 14, 16, 17, 18

    2023

  32. [40]

    Geo-bench: Toward foundation models for earth monitoring

    Alexandre Lacoste, Nils Lehmann, Pau Rodriguez, Evan Sherwin, Hannah Kerner, Bj¨orn L¨utjens, Jeremy Irvin, David Dao, Hamed Alemohammad, Alexandre Drouin, et al. Geo-bench: Toward foundation models for earth monitoring. NeurIPS, 2023. 7, 8

    2023

  33. [41]

    Pre- dicting gradient is better: Exploring self-supervised learning for sar atr with a joint-embedding predictive architecture

    Weijie Li, Wei Yang, Tianpeng Liu, Yuenan Hou, Yux- uan Li, Zhen Liu, Yongxiang Liu, and Li Liu. Pre- dicting gradient is better: Exploring self-supervised learning for sar atr with a joint-embedding predictive architecture. ISPRS Journal of Photogrammetry and Remote Sensing, 2024. 3

    2024

  34. [42]

    Masked angle-aware autoencoder for remote sensing images

    Zhihao Li, Biao Hou, Siteng Ma, Zitong Wu, Xianpeng Guo, Bo Ren, and Licheng Jiao. Masked angle-aware autoencoder for remote sensing images. arXiv preprint arXiv:2408.01946, 2024. 2

    2024 arXiv

  35. [43]

    Remoteclip: A vision language foundation model for remote sensing

    Fan Liu, Delong Chen, Zhangqingyun Guan, Xiaocong Zhou, Jiale Zhu, Qiaolin Ye, Liyong Fu, and Jun Zhou. Remoteclip: A vision language foundation model for remote sensing. IEEE Transactions on Geoscience and Remote Sensing, 2024. 8

    2024

  36. [44]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. In ICLR. 16

  37. [45]

    AI foundation models in remote sensing: A survey

    Siqi Lu, Junlin Guo, James R Zimmer-Dauphinee, Jor- dan M Nieusma, Xiao Wang, Parker VanValkenburgh, Steven A Wernke, and Yuankai Huo. AI foundation models in remote sensing: A survey. arXiv preprint arXiv:2408.03464, 2024. 1

    2024 arXiv

  38. [46]

    Se- mantic segmentation of crop type in Africa: A novel dataset and analysis of deep learning methods

    Rose M Rustowicz, Robin Cheong, Lijing Wang, Ste- fano Ermon, Marshall Burke, and David Lobell. Se- mantic segmentation of crop type in Africa: A novel dataset and analysis of deep learning methods. In CVPR Workshop EarthVision, 2019. 16

    2019

  39. [47]

    Seasonal contrast: Unsupervised pre-training from uncurated remote sens- ing data

    Oscar Manas, Alexandre Lacoste, Xavier Gir´o-i Nieto, David Vazquez, and Pau Rodriguez. Seasonal contrast: Unsupervised pre-training from uncurated remote sens- ing data. In ICCV, 2021. 2

    2021

  40. [48]

    Cross-sensor self-supervised training and alignment for remote sens- ing

    Valerio Marsocci and Nicolas Audebert. Cross-sensor self-supervised training and alignment for remote sens- ing. arXiv preprint arXiv:2405.09922, 2024. 5

    2024 arXiv

  41. [49]

    PANGAEA: A global and inclusive bench- mark for geospatial foundation models

    Valerio Marsocci, Yuru Jia, Georges Le Bellier, David Kerekes, Liang Zeng, Sebastian Hafner, Se- bastian Gerard, Eric Brune, Ritu Yadav, Ali Shibli, et al. PANGAEA: A global and inclusive bench- mark for geospatial foundation models. arXiv preprint arXiv:2412.04204, 2024. 7, 8

    2024 arXiv

  42. [50]

    Rethinking transformers pre- training for multi-spectral satellite imagery

    Mubashir Noman, Muzammal Naseer, Hisham Cholakkal, Rao Muhammad Anwer, Salman Khan, and Fahad Shahbaz Khan. Rethinking transformers pre- training for multi-spectral satellite imagery. In CVPR,

  43. [51]

    Unsupervised learn- ing of visual representations by solving jigsaw puzzles

    Mehdi Noroozi and Paolo Favaro. Unsupervised learn- ing of visual representations by solving jigsaw puzzles. In ECCV, 2016. 2

    2016

  44. [52]

    Rep- resentation learning with contrastive predictive coding

    Aaron van den Oord, Yazhe Li, and Oriol Vinyals. Rep- resentation learning with contrastive predictive coding. arXiv preprint arXiv:1807.03748, 2018. 4

    2018 arXiv

  45. [53]

    Dinov2: Learning robust visual fea- tures without supervision

    Maxime Oquab, Timoth´ee Darcet, Th´eo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fer- nandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, et al. Dinov2: Learning robust visual fea- tures without supervision. TLMR, 2023. 1, 2, 8, 15

    2023

  46. [54]

    Planted: A dataset for planted forest identification from multi-satellite time series

    Luis Miguel Pazos-Out ´on, Cristina Nader Vasconce- los, Anton Raichuk, Anurag Arnab, Dan Morris, and Maxim Neumann. Planted: A dataset for planted forest identification from multi-satellite time series. Inter- national Geoscience and Remote Sensing Symposium,

  47. [55]

    HLS foundation burnscars dataset, 2023

    Christopher Phillips, Sujit Roy, Kumar Ankur, and Rahul Ramachandran. HLS foundation burnscars dataset, 2023. 7, 8, 16, 17, 18

    2023

  48. [56]

    6, 7, 15, 16, 17, 18

  49. [57]

    A generalizable and acces- sible approach to machine learning with global satellite imagery

    Esther Rolf, Jonathan Proctor, Tamma Carleton, Ian Bolliger, Vaishaal Shankar, Miyabi Ishihara, Benjamin Recht, and Solomon Hsiang. A generalizable and acces- sible approach to machine learning with global satellite imagery. Nature communications, 2021. 15

    2021

  50. [58]

    Scale-MAE: A scale-aware masked autoencoder for multiscale geospatial representation learning

    Colorado J Reed, Ritwik Gupta, Shufan Li, Sarah Brockman, Christopher Funk, Brian Clipp, Kurt Keutzer, Salvatore Candido, Matt Uyttendaele, and Trevor Darrell. Scale-MAE: A scale-aware masked autoencoder for multiscale geospatial representation learning. In ICCV, 2023. 1, 2, 3...

    2023

  51. [59]

    SICKLE: A multi-sensor satellite imagery dataset anno- tated with multiple key cropping parameters

    Depanshu Sani, Sandeep Mahato, Sourabh Saini, Harsh Kumar Agarwal, Charu Chandra Devshali, Saket Anand, Gaurav Arora, and Thiagarajan Jayaraman. SICKLE: A multi-sensor satellite imagery dataset anno- tated with multiple key cropping parameters. In WACV,

  52. [60]

    Dos Santos, Maria Vakalopoulou, Ronny H¨ansch, Stine Hansen, Keiller Nogueira, Jonathan Prexl, and Devis Tuia

    Ribana Roscher, Marc Russwurm, Caroline Gevaert, Michael Kampffmeyer, Jefersson A. Dos Santos, Maria Vakalopoulou, Ronny H¨ansch, Stine Hansen, Keiller Nogueira, Jonathan Prexl, and Devis Tuia. Better, not just more: Data-centric machine learning for Earth observation. IEEE Ge...

    2024

  53. [61]

    Laion-5b: An open large-scale dataset for training next generation image-text models

    Christoph Schuhmann, Romain Beaumont, Richard Vencu, Cade Gordon, Ross Wightman, Mehdi Cherti, Theo Coombes, Aarush Katta, Clayton Mullis, Mitchell Wortsman, et al. Laion-5b: An open large-scale dataset for training next generation image-text models. In NeurIPS Dataset and ben...

    2022

  54. [62]

    TaxaBind: A uni- fied embedding space for ecological applications

    Srikumar Sastry, Subash Khanal, Aayush Dhakal, Adeel Ahmad, and Nathan Jacobs. TaxaBind: A uni- fied embedding space for ecological applications. In WACV, 2025. 2 11

    2025

  55. [63]

    Convo- lutional LSTM network: A machine learning approach for precipitation nowcasting

    Xingjian Shi, Zhourong Chen, Hao Wang, Dit-Yan Yeung, Wai-Kin Wong, and Wang-Chun Woo. Convo- lutional LSTM network: A machine learning approach for precipitation nowcasting. In NeurIPS, 2015. 16

    2015

  56. [64]

    Long short- term memory

    Hochreiter Sepp and Schmidhuber J¨urgen. Long short- term memory. Supervised sequence labelling with re- current neural networks, 2012. 16

    2012

  57. [65]

    Prithvi-EO-2.0: A versatile multi-temporal founda- tion model for earth observation applications

    Daniela Szwarcman, Sujit Roy, Paolo Fraccaro, Thorsteinn El´ı G´ıslason, Benedikt Blumenstiel, Rinki Ghosal, Pedro Henrique de Oliveira, Joao Lucas de Sousa Almeida, Rocco Sedona, Yanghui Kang, et al. Prithvi-EO-2.0: A versatile multi-temporal founda- tion model for earth obse...

    2024

  58. [66]

    SSSL4EO-l: Datasets and foundation models for Land- sat imagery

    Adam Stewart, Nils Lehmann, Isaac Corley, Yi Wang, Yi-Chia Chang, Nassim Ait Ait Ali Braham, Shradha Sehgal, Caleb Robinson, and Arindam Banerjee. SSSL4EO-l: Datasets and foundation models for Land- sat imagery. NeurIPS, 36, 2024. 15

    2024

  59. [67]

    Omni-scale CNNs: A simple and effective kernel size configuration for time series classification

    Wensi Tang, Guodong Long, Lu Liu, Tianyi Zhou, Michael Blumenstein, and Jing Jiang. Omni-scale CNNs: A simple and effective kernel size configuration for time series classification. In ICLR, 2021. 16

    2021

  60. [68]

    Cross-scale mae: A tale of multiscale exploitation in remote sensing

    Maofeng Tang, Andrei Cozma, Konstantinos Georgiou, and Hairong Qi. Cross-scale mae: A tale of multiscale exploitation in remote sensing. In NeurIPS, 2024. 2

    2024

  61. [69]

    Lightweight, pre-trained transformers for remote sensing timeseries

    Gabriel Tseng, Ivan Zvonkov, Mirali Purohit, David Rolnick, and Hannah Kerner. Lightweight, pre-trained transformers for remote sensing timeseries. arXiv preprint arXiv:2304.14065, 2023. 2, 15

    2023 arXiv

  62. [70]

    ViTs for SITS: Vision transformers for satellite image time series

    Michail Tarasiou, Erik Chavez, and Stefanos Zafeiriou. ViTs for SITS: Vision transformers for satellite image time series. In CVPR, 2023. 15

    2023

  63. [71]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. In NeurIPS, 2017. 16

    2017

  64. [72]

    CROCO: Cross- modal contrastive learning for localization of Earth observation data

    Wei-Hsin Tseng, Ho `ang- ˆAn L ˆe, Alexandre Boulch, S´ebastien Lef`evre, and Dirk Tiede. CROCO: Cross- modal contrastive learning for localization of Earth observation data. ISPRS Annals of the Photogramme- try, Remote Sensing and Spatial Information Sciences,

  65. [73]

    Extracting and composing robust features with denoising autoencoders

    Pascal Vincent, Hugo Larochelle, Yoshua Bengio, and Pierre-Antoine Manzagol. Extracting and composing robust features with denoising autoencoders. In ICML,

  66. [74]

    Satel- lite image time series semantic change detection: Novel architecture and analysis of domain shift

    Elliot Vincent, Jean Ponce, and Mathieu Aubry. Satel- lite image time series semantic change detection: Novel architecture and analysis of domain shift. arXiv preprint arXiv:2407.07616, 2024. 16

    2024 arXiv

  67. [75]

    Timesen2crop: A million labeled samples dataset of Sentinel 2 image time series for crop-type classifica- tion

    Giulio Weikmann, Claudia Paris, and Lorenzo Bruz- zone. Timesen2crop: A million labeled samples dataset of Sentinel 2 image time series for crop-type classifica- tion. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021. 7, 16, 17, 18

    2021

  68. [76]

    SSL4EO-S12: A large-scale multi-modal, multi- temporal dataset for self-supervised learning in Earth observation

    Yi Wang, Nassim Ait Ali Braham, Zhitong Xiong, Chenying Liu, Conrad M Albrecht, and Xiao Xiang Zhu. SSL4EO-S12: A large-scale multi-modal, multi- temporal dataset for self-supervised learning in Earth observation. IEEE Geoscience and Remote Sensing Magazine, 2023. 1

    2023

  69. [77]

    Neural plasticity-inspired foundation model for observing the Earth crossing modalities

    Zhitong Xiong, Yi Wang, Fahong Zhang, Adam J Stew- art, Jo ¨elle Hanna, Damian Borth, Ioannis Papoutsis, Bertrand Le Saux, Gustau Camps-Valls, and Xiao Xi- ang Zhu. Neural plasticity-inspired foundation model for observing the Earth crossing modalities. arXiv preprint arXiv:24...

    2024

  70. [78]

    Zooming out on zooming in: Advancing super- resolution for remote sensing, 2023

    Piper Wolters, Favyen Bastani, and Aniruddha Kemb- havi. Zooming out on zooming in: Advancing super- resolution for remote sensing, 2023. 17, 18

    2023

  71. [79]

    SITS-Former: A pre-trained spatio- spectral-temporal representation model for sentinel- 2 time series classification

    Yuan Yuan, Lei Lin, Qingshan Liu, Renlong Hang, and Zeng-Guang Zhou. SITS-Former: A pre-trained spatio- spectral-temporal representation model for sentinel- 2 time series classification. International Journal of Applied Earth Observation and Geoinformation, 2022. 2

    2022

  72. [80]

    Masked image modeling with denoising contrast

    Kun Yi, Yixiao Ge, Xiaotong Li, Shusheng Yang, Dian Li, Jianping Wu, Ying Shan, and Xiaohu Qie. Masked image modeling with denoising contrast. In ICLR,

  73. [82]

    So2Sat LCZ42: A benchmark dataset for global local climate zones classification

    Xiao Xiang Zhu, Jingliang Hu, Chunping Qiu, Yilei Shi, Jian Kang, Lichao Mou, Hossein Bagheri, Matthias H¨aberle, Yuansheng Hua, Rong Huang, et al. So2Sat LCZ42: A benchmark dataset for global local climate zones classification. arXiv preprint arXiv:1912.12171, 2019. 7, 16, 17...

    1912 arXiv

  74. [83]

    We used a LinearWarmupCosineAnnealingLR [1] for classifi- cation and ReduceLROnPlateau [2] scheduler for pre- training and segmentation

    optimizer with a learning rate of 5 × 10−5 for all our experiments (pretraining and fine-tuning). We used a LinearWarmupCosineAnnealingLR [1] for classifi- cation and ReduceLROnPlateau [2] scheduler for pre- training and segmentation. We set he contrastive temperatureγ to 0.1 ...

  75. [2024]

    [Online; accessed 01-Sept-2024]. 17, 18

    2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.