pith. sign in

arxiv: 2507.12590 · v2 · submitted 2025-07-16 · 💻 cs.CV · cs.LG

From Time-series Generation, Model Selection to Transfer Learning: A Comparative Review of Pixel-wise Approaches for Large-scale Crop Mapping

Judy Long , Tao Liu , Sean Alexander Woznicki , Miljana Markovi\'c , Oskar Marko , Molly Sears This is my paper

Pith reviewed 2026-05-19 03:52 UTC · model grok-4.3

classification 💻 cs.CV cs.LG
keywords crop mappingremote sensingpixel-wise classificationtransfer learningtime series preprocessingtransformer modelssupervised learningdomain adaptation
0
0 comments X

The pith

Fine-scale time-series preprocessing with Transformer models delivers top performance for large-scale crop mapping in both supervised and transfer settings.

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

The paper reviews and experimentally compares methods for preparing satellite time series and selecting models to map crop types at the pixel level over large areas. It tests six preprocessing approaches and eleven classification models on data from five agricultural sites using Landsat imagery and trusted labels. The central finding is that preprocessing with fine time intervals combined with Transformer models achieves the best results whether training directly on the target area or adapting from other regions. The work also shows that the best strategy shifts with the amount of labeled data available, favoring supervised methods when labels are plentiful and transfer learning when they are scarce.

Core claim

Through systematic experiments, the study demonstrates that fine-scale interval preprocessing paired with Transformer models consistently delivered optimal performance for both supervised and transferable workflows across the tested sites. Random forest models offer rapid training with competitive accuracy in conventional supervised learning and direct transfer to similar domains. Transfer learning techniques, particularly unsupervised domain adaptation for homogeneous classes and fine-tuning for diverse scenarios, enhance adaptability, with the choice depending on the availability of labeled samples.

What carries the argument

Comparative evaluation of time-series generation methods and pixel-wise classification models for crop type identification from multi-temporal remote sensing data.

If this is right

  • Supervised training with sufficient samples yields more accurate and generalizable crop maps than transfer approaches.
  • Matching the transfer learning technique to the degree of domain shift improves results when labeled data is limited.
  • Random forest remains a practical choice for quick implementation in similar domain transfers.
  • Workflow performance varies with sample size, variable combinations, and crop class homogeneity.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The results imply that temporal resolution in preprocessing may be more critical than model complexity in many remote sensing classification tasks.
  • Extending the approach to other land cover types or different satellite sensors could test the robustness of these optimal configurations.
  • Practitioners might prioritize acquiring high-quality labels from a few representative sites over broad but sparse coverage.

Load-bearing premise

The five agricultural sites along with CDL trusted pixels and field surveys supply representative and accurate ground-truth labels that hold for conditions outside the tested regions and data sources.

What would settle it

Applying the identified optimal preprocessing and models to an additional agricultural area with substantially different crop mixes, climate, or imaging conditions and finding markedly reduced accuracy would challenge the claim of consistent optimality.

Figures

Figures reproduced from arXiv: 2507.12590 by Judy Long, Miljana Markovi\'c, Molly Sears, Oskar Marko, Sean Alexander Woznicki, Tao Liu.

Figure 1
Figure 1. Figure 1: From 2010 to 2024, the number of studies integrating remote sensing with crop mapping increased by approximately 20% annually. From 2016 onward, the use of transfer learning in this field grew at a geometric mean annual rate of 81%. Data were retrieved from Google Scholar using two sets of search terms: (a) "Remote Sensing" AND "Crop Mapping" and (b) "Remote Sensing" AND "Crop Mapping" AND "Transfer Learni… view at source ↗
Figure 2
Figure 2. Figure 2: Geographic distribution of five study sites. Sites A, B, and C are located within the U.S. Corn and Soybean Belt and serve as both source and target sites in temporal and sensor transfer scenarios, denoted by blue borders. In the spatial transfer scenario, Site A is designated as the source site, while Sites B, D, and E are the target sites, denoted by orange borders. Site D lies in a transitional zone bet… view at source ↗
Figure 3
Figure 3. Figure 3: Geographic identifiers, valid pixel distribution, time series length, and crop-specific sample composition for each study site from April 1 to October 1, 2023. The footprint location column lists the geographic identifiers for each site, including Landsat path/row for U.S. regions and the province name for the Serbian site. Valid pixel count refers to the number of pixels with reliable surface reflectance … view at source ↗
Figure 4
Figure 4. Figure 4: Experimental design overview for evaluating pixel-wise crop mapping workflows. Crop maps were produced either via fully supervised learning or transfer learning. The supervised workflow comprises sequential modules: (1) time series reconstruction, (2) sample generation, (3) model selection, and (4) crop map prediction. Within each module, multiple widely used approaches were evaluated to identify optimal p… view at source ↗
Figure 5
Figure 5. Figure 5: Six image-wide preprocessing methods are exemplified using NIR reflectance from a trusted sample at Site A. The grey background represents the target crop growing season defined in 2023. Observation (dots) selection depends on whether reconstruction relies on native path and row composites (raw methods; (a) and (b)) or all available observations (linear resampling methods; (c)-(f)). The QA_PIXEL band was u… view at source ↗
Figure 6
Figure 6. Figure 6: Architecture illustration of pixel-based models. (a) Random Forest. (b) Transformer. (c) Single-layer RNN architecture (RNN, LSTM, and GRU). (d) Bidirectional RNN architecture (Bi-RNN, Bi-LSTM, and Bi-GRU). (e) Bidirectional RNN with an attention layer architecture (AtBi-RNN, AtBi-LSTM, and AtBi-GRU). Abbreviation: DT, decision tree model. band was counted as separable if its inter-class distance exceeded … view at source ↗
Figure 7
Figure 7. Figure 7: Overview of the DANN framework used for unsupervised domain adaptation. The model processes time series from both source and target domains using a shared deep feature extractor. The resulting feature representations are passed to two branches: a label predictor trained on source labels and a domain classifier trained to distinguish between source and target samples. A gradient reversal layer connects the … view at source ↗
Figure 8
Figure 8. Figure 8: Processed time series of six Landsat 8 optical bands and two Sentinel-1 polarization bands for Site A in 2023. The x-axis shows the time in the format of either DOY or time steps, and the y-axis represents the reflectance value. (a) Raw reflectance values from the growing season. (b) and (g) Weighted WE smoother applied to raw observations. (c) 7-day linear resampling. (d) 30-day linear resampling. (e) WE-… view at source ↗
Figure 9
Figure 9. Figure 9: Classification accuracy and computational cost of six preprocessing methods and eleven pixel-wise classification models at three Corn-Soybean Belt sites. (a) Overall accuracy of six preprocessing methods combined with five representative models, sorted by ascending average accuracy. Three finer-scale linear resampling-based methods show consistent superiority. (b) Overall accuracy of eleven models paired w… view at source ↗
Figure 10
Figure 10. Figure 10: Six sample sizes and four variable combinations were evaluated on optimal RF and Transformer workflows. (a) Boxplot of 10-fold cross-validation overall accuracies. Accuracy improves and becomes more stable as the training sample size increases, though gains diminish at larger sample sizes. Incorporating all complementary variables enhances RF classification performance relative to using optical bands alon… view at source ↗
Figure 11
Figure 11. Figure 11: Comparison of model transferability across temporal, sensor, and spatial domains. Each grouped bar plot summarizes the overall accuracy of eleven pixel-wise classification models combined with three optimal preprocessing methods. Colored bars represent the OA of transferred pre-trained models, while gray bars indicate the upper-bound accuracies achievable in the target domain via supervised learning, assu… view at source ↗
Figure 12
Figure 12. Figure 12: Confusion matrices for target-site test sets under (a) temporal transfer at Site C and (b) spatial transfer to Site E, comparing single-source, multi-source, and fully supervised models. Fully supervised Transformer models were trained on 9,000 labeled samples from each target site. Class labels are corn (C), soybean (S), and "other" (O). Values represent row￾normalized proportions; higher diagonal values… view at source ↗
Figure 13
Figure 13. Figure 13: Comparison of per-class producer’s accuracy for direct transfer, DANN, and complete fine-tuning workflows across spatial transfer scenarios. Rows represent spatial transfer pairs with varying levels of domain shifts (A to B: low, A to D: moderate, A to E: challenging.) Within each panel, DL models are ordered left-to-right: 1) RNN, 2) Bi-RNN, 3) AtBi-RNN, 4) GRU, 5) Bi-GRU, 6) AtBi-GRU, 7) LSTM, 8) Bi-LST… view at source ↗
Figure 14
Figure 14. Figure 14: Difference maps and prediction maps at Site B, generated using direct transfer, DANN-based transfer learning, and fully supervised learning workflows (zoomed area covers ~225 km2 ). Under low domain shift, both direct transfer and DANN-based models can achieve high-accuracy crop mapping without requiring labeled data from the target site (over 90% OA). The reference map was CDL in 2023, with 85% to 95% pr… view at source ↗
Figure 15
Figure 15. Figure 15: Training loss and validation accuracy curves for four fine-tuning strategies, including basic complete fine-tuning (R1), class-weighted sampling and loss (R2), balanced subset undersampling (R3), and two-stage fine-tuning (R4). The pre-trained models used training samples from Sites A and D [PITH_FULL_IMAGE:figures/full_fig_p027_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Per-class producer’s accuracy from single-source and multi-source domain fine-tuning checkpoints. Multi-source domain fine-tuning did not yield clear advantages over single-source approaches under large domain shifts, consistent with the findings in direct transfer results. The average accuracy was calculated from 8 out of 10 cross-validation folds, excluding the lowest and highest fold results to minimiz… view at source ↗
Figure 17
Figure 17. Figure 17: Difference maps and prediction maps at Site E, generated using direct transfer, fine-tuning R3-based transfer learning, and fully supervised learning workflows (zoomed area covers ~225 km2 ). The reference map was obtained from the BioSense Institute with a 10-m spatial resolution and an overall accuracy over 90% (Pandžić et al., 2024; Živaljević et al., 2024). The difference map shows the areas of mismat… view at source ↗
Figure 18
Figure 18. Figure 18: Workflow choice depends primarily on labeled sample availability, while transfer learning strategy selection is further influenced by the magnitude of domain shift. The sample size reflects the availability of labeled data at the target site, where the threshold between limited and sufficient data varies by spatial extent, crop types, and class balance. Domain shift reflects the degree of differences in l… view at source ↗
Figure 19
Figure 19. Figure 19: Overall accuracy from 10-fold cross-validation for RF and Transformer models across three preprocessing methods and two sample types at Sites A, B, and C. Each violin plot summarizes 30 accuracy values per model, comprising results from three preprocessing strategies (7-day linear resampling, WE-smoothed 7-day linear resampling, and phenological peak), each evaluated with 10 folds. Color clusters highligh… view at source ↗
Figure 20
Figure 20. Figure 20: Comparison of model predictions using original CDL versus trusted samples for a 500-by-500-pixel subset (~225 km2 ) at Site A. RF trained on original CDL and Transformer trained on trusted samples are shown alongside the 2023 CDL reference map. Transformer predictions show clearer, more coherent patterns, especially at field boundaries. Abbreviations: CDL, 2023 Cropland Data Layer; RF, Random Forest model… view at source ↗
Figure 21
Figure 21. Figure 21: NDVI and MSI time series curves for corn and soybean across spatial, temporal, and sensor transfer scenarios, indicating seasonal changes in vegetation greenness and water stress. Shaded areas represent one standard deviation for all scenarios except Site E, where the shaded area represents the 95th percentile. Overall, both crops reach peak greenness (NDVI) around July, coinciding with decreasing water s… view at source ↗
Figure 22
Figure 22. Figure 22: Comparison of transfer learning approaches across spatial domains in 2023. The grouped bar plot summarizes the overall accuracy of eleven pixel-wise classification models combined with three preprocessing methods. Grey bars indicate the test accuracy of the upper-bound model trained from scratch for each target site, while the colored bars indicate the test accuracy of the direct-transfer model. The arrow… view at source ↗
read the original abstract

Crop mapping involves identifying and classifying crop types using spatial data, primarily derived from remote sensing imagery. This study presents the first comprehensive review of large-scale, pixel-wise crop mapping workflows, encompassing both conventional supervised methods and emerging transfer learning approaches. To identify the optimal time-series generation approaches and supervised crop mapping models, we conducted systematic experiments, comparing six widely adopted satellite image-based preprocessing methods, alongside eleven supervised pixel-wise classification models. Additionally, we assessed the synergistic impact of varied training sample sizes and variable combinations. Moreover, we identified optimal transfer learning techniques for different magnitudes of domain shift. The evaluation of optimal methods was conducted across five diverse agricultural sites. Landsat 8 served as the primary satellite data source. Labels come from CDL trusted pixels and field surveys. Our findings reveal three key insights. First, fine-scale interval preprocessing paired with Transformer models consistently delivered optimal performance for both supervised and transferable workflows. RF offered rapid training and competitive performance in conventional supervised learning and direct transfer to similar domains. Second, transfer learning techniques enhanced workflow adaptability, with UDA being effective for homogeneous crop classes while fine-tuning remains robust across diverse scenarios. Finally, workflow choice depends heavily on the availability of labeled samples. With a sufficient sample size, supervised training typically delivers more accurate and generalizable results. Below a certain threshold, transfer learning that matches the level of domain shift is a viable alternative to achieve crop mapping. All code is publicly available to encourage reproducibility practice.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript presents the first comprehensive review and experimental comparison of pixel-wise crop mapping workflows using Landsat 8 time-series data. It evaluates six satellite image preprocessing methods for time-series generation and eleven supervised classification models across five agricultural sites, with ground truth from CDL trusted pixels and field surveys. The study further examines the impact of training sample sizes, variable combinations, and transfer learning techniques (including UDA and fine-tuning) for varying domain shifts. The central claim is that fine-scale interval preprocessing combined with Transformer models consistently yields optimal performance for both supervised and transferable workflows, while RF is competitive for rapid training in similar domains; workflow choice depends on labeled sample availability, with supervised training preferred above a threshold and matched transfer learning below it. All code is released publicly.

Significance. If the empirical rankings hold under more rigorous validation, the work provides actionable guidance for large-scale crop mapping practitioners by identifying effective preprocessing-model combinations and transfer strategies. The public code release is a clear strength that supports reproducibility in remote sensing and agricultural CV applications.

major comments (2)
  1. [Abstract and Results] Abstract and Results sections: The headline claim that fine-scale interval preprocessing paired with Transformer models 'consistently delivered optimal performance' for supervised and transferable workflows is not accompanied by statistical significance tests (e.g., McNemar or paired t-tests on performance deltas), error bars from multiple random seeds, or hyperparameter fairness controls. This undermines confidence in the method rankings across the six preprocessing pipelines and eleven models.
  2. [Methods] Methods (Ground Truth and Experimental Setup): The evaluation depends on CDL trusted pixels plus field surveys without any reported sensitivity analysis or uncertainty propagation for known CDL label noise (typically 5-15% commission/omission errors). If these errors are non-uniform across classes or sites, the relative optimality of preprocessing and models could shift, making this a load-bearing concern for the cross-site claims.
minor comments (2)
  1. [Abstract] Abstract: Explicitly enumerate the six preprocessing methods and eleven models to improve immediate readability of the experimental scope.
  2. [Throughout] Notation and figures: Ensure consistent terminology for transfer learning components (e.g., UDA vs. fine-tuning) and improve clarity of any performance tables or plots showing cross-site results.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thoughtful and constructive comments. We address each major point below and outline specific revisions to strengthen the empirical claims and robustness of the analysis.

read point-by-point responses

  1. Referee: [Abstract and Results] Abstract and Results sections: The headline claim that fine-scale interval preprocessing paired with Transformer models 'consistently delivered optimal performance' for supervised and transferable workflows is not accompanied by statistical significance tests (e.g., McNemar or paired t-tests on performance deltas), error bars from multiple random seeds, or hyperparameter fairness controls. This undermines confidence in the method rankings across the six preprocessing pipelines and eleven models.

    Authors: We agree that formal statistical testing and variability reporting would increase confidence in the reported rankings. In the revised version we will (i) run each model with three independent random seeds and report mean performance with standard deviation error bars, (ii) apply paired t-tests (or McNemar tests for classification accuracy) to the key performance deltas between the top-ranked and runner-up pipelines, and (iii) document that hyper-parameter search budgets were equalized across models using the same Optuna configuration and early-stopping protocol. These additions will be placed in a new subsection of Results and referenced in the Abstract. revision: yes

  2. Referee: [Methods] Methods (Ground Truth and Experimental Setup): The evaluation depends on CDL trusted pixels plus field surveys without any reported sensitivity analysis or uncertainty propagation for known CDL label noise (typically 5-15% commission/omission errors). If these errors are non-uniform across classes or sites, the relative optimality of preprocessing and models could shift, making this a load-bearing concern for the cross-site claims.

    Authors: We acknowledge that CDL label noise constitutes a potential confounding factor. While the 'trusted pixel' filtering already removes low-confidence CDL labels, we did not quantify its residual impact. In revision we will add a sensitivity experiment that injects controlled label noise (5 %, 10 %, 15 %) at both class-balanced and class-imbalanced rates, re-train the top three pipelines on each perturbed label set, and report how the relative ordering of preprocessing–model combinations changes across the five sites. Results will appear in a new subsection of Methods and a supplementary figure. revision: yes

Circularity Check

0 steps flagged

No circularity: purely empirical comparison with held-out evaluations

full rationale

The paper reports results from systematic experiments comparing six preprocessing pipelines and eleven models across five agricultural sites using Landsat 8 imagery and CDL/field-survey labels. No mathematical derivations, first-principles predictions, or equations are present that could reduce to fitted parameters or self-citations by construction. All optimality claims rest on direct performance measurements on held-out sites rather than any definitional or fitted-input loop, satisfying the self-contained benchmark criterion.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is an empirical benchmarking study rather than a theoretical derivation. No mathematical free parameters or invented physical entities are introduced. Model and preprocessing selections are drawn from existing literature and tested experimentally.

pith-pipeline@v0.9.0 · 5816 in / 1144 out tokens · 43519 ms · 2026-05-19T03:52:09.056832+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

115 extracted references · 115 canonical work pages · 5 internal anchors

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

    work page

  2. [2]

    , author Kumar, A

    author Aleissaee, A.A. , author Kumar, A. , author Anwer, R.M. , author Khan, S. , author Cholakkal, H. , author Xia, G.S. , author Khan, F.S. , year 2023 . title Transformers in remote sensing: A survey . journal Remote Sensing volume 15 , pages 1860

    work page 2023

  3. [3]

    Neural Machine Translation by Jointly Learning to Align and Translate

    author Bahdanau, D. , year 2014 . title Neural machine translation by jointly learning to align and translate . journal arXiv preprint arXiv:1409.0473

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  4. [4]

    , author Yang, Z

    author Boryan, C. , author Yang, Z. , author Mueller, R. , author Craig, M. , year 2011 . title Monitoring us agriculture: the us department of agriculture, national agricultural statistics service, cropland data layer program . journal Geocarto International volume 26 , pages 341--358

    work page 2011

  5. [5]

    , author Mueller, R

    author Boudiaf, M. , author Mueller, R. , author Ben Ayed, I. , author Bertinetto, L. , year 2022 . title Parameter-free online test-time adaptation , in: booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. pages 8344--8353

    work page 2022

  6. [6]

    Random forests

    author Breiman, L. , year 2001 . title Random forests . journal Mach. Learn. volume 45 , pages 5–32 . https://doi.org/10.1023/A:1010933404324, :10.1023/a:1010933404324

    work page doi:10.1023/a:1010933404324 2001

  7. [7]

    , author Guan, K

    author Cai, Y. , author Guan, K. , author Peng, J. , author Wang, S. , author Seifert, C. , author Wardlow, B. , author Li, Z. , year 2018 . title A high-performance and in-season classification system of field-level crop types using time-series landsat data and a machine learning approach . journal Remote Sensing of Environment volume 210 , pages 35--47 ...

    work page doi:10.1016/j.rse.2018.02.045 2018

  8. [8]

    , year 1997

    author Caruana, R. , year 1997 . title Multitask learning . journal Machine learning volume 28 , pages 41--75

    work page 1997

  9. [9]

    , author Chen, S

    author Chen, J. , author Chen, S. , author Fu, R. , author Li, D. , author Jiang, H. , author Wang, C. , author Peng, Y. , author Jia, K. , author Hicks, B.J. , year 2022 . title Remote sensing big data for water environment monitoring: current status, challenges, and future prospects . journal Earth's Future volume 10 , pages e2021EF002289

    work page 2022

  10. [10]

    , author Cao, R

    author Chen, Y. , author Cao, R. , author Chen, J. , author Liu, L. , author Matsushita, B. , year 2021 . title A practical approach to reconstruct high-quality landsat ndvi time-series data by gap filling and the savitzky–golay filter . journal ISPRS Journal of Photogrammetry and Remote Sensing volume 180 , pages 174--190 . https://www.sciencedirect.com/...

    work page doi:10.1016/j.isprsjprs.2021.08.015 2021

  11. [11]

    , author Song, X

    author Chen, Y. , author Song, X. , author Wang, S. , author Huang, J. , author Mansaray, L.R. , year 2016 . title Impacts of spatial heterogeneity on crop area mapping in canada using modis data . journal ISPRS Journal of Photogrammetry and Remote Sensing volume 119 , pages 451--461

    work page 2016

  12. [12]

    On the Properties of Neural Machine Translation: Encoder-Decoder Approaches

    author Cho, K. , year 2014 . title On the properties of neural machine translation: Encoder-decoder approaches . journal arXiv preprint arXiv:1409.1259

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  13. [13]

    Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling

    author Chung, J. , author Gulcehre, C. , author Cho, K. , author Bengio, Y. , year 2014 . title Empirical evaluation of gated recurrent neural networks on sequence modeling . journal arXiv [cs.NE] http://arxiv.org/abs/1412.3555

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  14. [14]

    , author Belgiu, M

    author Csillik, O. , author Belgiu, M. , author Asner, G.P. , author Kelly, M. , year 2019 . title Object-based time-constrained dynamic time warping classification of crops using sentinel-2 . journal Remote sensing volume 11 , pages 1257

    work page 2019

  15. [15]

    , author Lavhale, P

    author Diwate, P. , author Lavhale, P. , author Pande, C.B. , author Sammen, S.S. , author Refadah, S.S. , author Khan, M.Y.A. , author Elkhrachy, I. , author Salem, A. , year 2025 . title Evaluating flood dynamics and effects in nagpur city using remote sensing and shannon’s entropy analysis . journal Scientific Reports volume 15 , pages 4900

    work page 2025

  16. [16]

    , author Pesendorfer, V

    author Eilers, P.H. , author Pesendorfer, V. , author Bonifacio, R. , year 2017 . title Automatic smoothing of remote sensing data , in: booktitle 2017 9th International Workshop on the Analysis of Multitemporal Remote Sensing Images (MultiTemp) , pp. pages 1--3 . :10.1109/Multi-Temp.2017.8076705

    work page doi:10.1109/multi-temp.2017.8076705 2017

  17. [17]

    , year 1990

    author Elman, J.L. , year 1990 . title Finding structure in time . journal Cognitive science volume 14 , pages 179--211

    work page 1990

  18. [18]

    title Sentinel-1 sar grd product

    author ESA , year 2025 . title Sentinel-1 sar grd product . howpublished https://browser.dataspace.copernicus.eu/ . note Accessed: 2025-02-05

    work page 2025

  19. [19]

    , author Xia, L

    author Fan, L. , author Xia, L. , author Yang, J. , author Sun, X. , author Wu, S. , author Qiu, B. , author Chen, J. , author Wu, W. , author Yang, P. , year 2024 . title A temporal-spatial deep learning network for winter wheat mapping using time-series sentinel-2 imagery . journal ISPRS Journal of Photogrammetry and Remote Sensing volume 214 , pages 48--64

    work page 2024

  20. [20]

    , author Gao, M

    author Feng, F. , author Gao, M. , author Liu, R. , author Yao, S. , author Yang, G. , year 2023 . title A deep learning framework for crop mapping with reconstructed sentinel-2 time series images . journal Computers and Electronics in Agriculture volume 213 , pages 108227 . https://www.sciencedirect.com/science/article/pii/S0168169923006154, :https://doi...

    work page doi:10.1016/j.compag.2023.108227 2023

  21. [21]

    , author Rollin, P

    author Fisette, T. , author Rollin, P. , author Aly, Z. , author Campbell, L. , author Daneshfar, B. , author Filyer, P. , author Smith, A. , author Davidson, A. , author Shang, J. , author Jarvis, I. , year 2013 . title Aafc annual crop inventory , in: booktitle 2013 Second international conference on agro-geoinformatics (agro-geoinformatics) , organizat...

    work page 2013

  22. [22]

    , author Shen, R

    author Fu, Y. , author Shen, R. , author Song, C. , author Dong, J. , author Han, W. , author Ye, T. , author Yuan, W. , year 2023 . title Exploring the effects of training samples on the accuracy of crop mapping with machine learning algorithm . journal Science of Remote Sensing volume 7 , pages 100081

    work page 2023

  23. [23]

    , author Vatsavai, R.R

    author Gadiraju, K.K. , author Vatsavai, R.R. , year 2020 . title Comparative analysis of deep transfer learning performance on crop classification , in: booktitle Proceedings of the 9th ACM SIGSPATIAL International Workshop on Analytics for Big Geospatial Data , pp. pages 1--8

    work page 2020

  24. [24]

    , author Lempitsky, V

    author Ganin, Y. , author Lempitsky, V. , year 2015 . title Unsupervised domain adaptation by backpropagation , in: booktitle International conference on machine learning , organization PMLR . pp. pages 1180--1189

    work page 2015

  25. [25]

    Domain-Adversarial Training of Neural Networks

    author Ganin, Y. , author Ustinova, E. , author Ajakan, H. , author Germain, P. , author Larochelle, H. , author Laviolette, F. , author Marchand, M. , author Lempitsky, V. , year 2016 . title Domain-adversarial training of neural networks . journal arXiv [stat.ML] http://arxiv.org/abs/1505.07818

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  26. [26]

    , year 1996

    author Gao, B.C. , year 1996 . title Ndwi—a normalized difference water index for remote sensing of vegetation liquid water from space . journal Remote sensing of environment volume 58 , pages 257--266

    work page 1996

  27. [27]

    , author Markovic, S

    author Gavrilov, M. , author Markovic, S. , author Jarad, A. , author Korac, V. , year 2015 . title The analysis temperature trends in vojvodina (serbia) from 1949 to 2006 . journal Thermal Science volume 19 , pages 339--350 . :10.2298/TSCI150207062G

    work page doi:10.2298/tsci150207062g 2015

  28. [28]

    , author Gritz † , Y

    author Gitelson, A.A. , author Gritz † , Y. , author Merzlyak, M.N. , year 2003 . title Relationships between leaf chlorophyll content and spectral reflectance and algorithms for non-destructive chlorophyll assessment in higher plant leaves . journal Journal of Plant Physiology volume 160 , pages 271--282 . https://www.sciencedirect.com/science/article/pi...

    work page doi:10.1078/0176-1617-00887 2003

  29. [29]

    , author Hancher, M

    author Gorelick, N. , author Hancher, M. , author Dixon, M. , author Ilyushchenko, S. , author Thau, D. , author Moore, R. , year 2017 . title Google earth engine: Planetary-scale geospatial analysis for everyone . journal Remote sensing of Environment volume 202 , pages 18--27

    work page 2017

  30. [31]

    , author Nendel, C

    author Griffiths, P. , author Nendel, C. , author Hostert, P. , year 2019 b. title Intra-annual reflectance composites from sentinel-2 and landsat for national-scale crop and land cover mapping . journal Remote Sensing of Environment volume 220 , pages 135--151 . https://www.sciencedirect.com/science/article/pii/S0034425718304851, :https://doi.org/10.1016...

    work page doi:10.1016/j.rse.2018.10.031 2019

  31. [32]

    , author Clark, M.L

    author Hakkenberg, C.R. , author Clark, M.L. , author Bailey, T. , author Burns, P. , author Goetz, S.J. , year 2024 . title Ladder fuels rather than canopy volumes consistently predict wildfire severity even in extreme topographic-weather conditions . journal Communications Earth & Environment volume 5 , pages 721

    work page 2024

  32. [33]

    Long short- term memory,

    author Hochreiter, S. , author Schmidhuber, J. , year 1997 . title Long short-term memory . journal Neural Comput. volume 9 , pages 1735–1780 . https://doi.org/10.1162/neco.1997.9.8.1735, :10.1162/neco.1997.9.8.1735

    work page doi:10.1162/neco.1997.9.8.1735 1997

  33. [34]

    , author Liu, H

    author Huete, A. , author Liu, H. , author Batchily, K. , author Van Leeuwen, W. , year 1997 . title A comparison of vegetation indices over a global set of tm images for eos-modis . journal Remote sensing of environment volume 59 , pages 440--451

    work page 1997

  34. [35]

    title Agricultural Water Management in Serbia

    author IWSU , year 2023 . title Agricultural Water Management in Serbia . type Technical Report . IWSU. https://www.iwsu.com.tr/assets/pdf/technicalreports/Tec_Rep_IV_Ag_Water_Mng_Serbia

    work page 2023

  35. [36]

    , author Shu, Y

    author Jiang, J. , author Shu, Y. , author Wang, J. , author Long, M. , year 2022 . title Transferability in deep learning: A survey . http://arxiv.org/abs/2201.05867 arXiv:2201.05867

    work page arXiv 2022

  36. [37]

    , author Pradhan, B

    author Joshi, A. , author Pradhan, B. , author Gite, S. , author Chakraborty, S. , year 2023 . title Remote-sensing data and deep-learning techniques in crop mapping and yield prediction: A systematic review . journal Remote Sensing volume 15 . https://www.mdpi.com/2072-4292/15/8/2014, :10.3390/rs15082014

    work page doi:10.3390/rs15082014 2023

  37. [38]

    Chandrasekar, M

    author K. Chandrasekar, M. V. R. Sesha Sai, P.S.R. , author Dwevedi, R.S. , year 2010 . title Land surface water index (lswi) response to rainfall and ndvi using the modis vegetation index product . journal International Journal of Remote Sensing volume 31 , pages 3987--4005 . https://doi.org/10.1080/01431160802575653, :10.1080/01431160802575653, http://a...

    work page doi:10.1080/01431160802575653 2010

  38. [39]

    , author Sahajpal, R

    author Kerner, H.R. , author Sahajpal, R. , author Pai, D.B. , author Skakun, S. , author Puricelli, E. , author Hosseini, M. , author Meyer, S. , author Becker-Reshef, I. , year 2022 . title Phenological normalization can improve in-season classification of maize and soybean: A case study in the central us corn belt . journal Science of Remote Sensing vo...

    work page 2022

  39. [40]

    , author Pham, H

    author Khaki, S. , author Pham, H. , author Wang, L. , year 2021 . title Simultaneous corn and soybean yield prediction from remote sensing data using deep transfer learning . journal Scientific Reports volume 11 , pages 11132

    work page 2021

  40. [41]

    , author Matin, M.A

    author Khanal, N. , author Matin, M.A. , author Uddin, K. , author Poortinga, A. , author Chishtie, F. , author Tenneson, K. , author Saah, D. , year 2020 . title A comparison of three temporal smoothing algorithms to improve land cover classification: A case study from nepal . journal Remote Sensing volume 12 , pages 2888 . https://www.mdpi.com/2072-4292...

    work page 2020

  41. [42]

    , author Shariff, A.R.M

    author Kok, Z.H. , author Shariff, A.R.M. , author Alfatni, M.S.M. , author Khairunniza-Bejo, S. , year 2021 . title Support vector machine in precision agriculture: a review . journal Computers and Electronics in Agriculture volume 191 , pages 106546

    work page 2021

  42. [43]

    , author Kumar, J

    author Konduri, V.S. , author Kumar, J. , author Hargrove, W.W. , author Hoffman, F.M. , author Ganguly, A.R. , year 2020 . title Mapping crops within the growing season across the united states . journal Remote Sensing of Environment volume 251 , pages 112048

    work page 2020

  43. [44]

    , author Grieser, J

    author Kottek, M. , author Grieser, J. , author Beck, C. , author Rudolf, B. , author Rubel, F. , year 2006 . title World map of the köppen-geiger climate classification updated . journal Meteorologische Zeitschrift volume 15 , pages 259--263 . :10.1127/0941-2948/2006/0130

    work page doi:10.1127/0941-2948/2006/0130 2006

  44. [45]

    , author Skakun, S

    author Kussul, N. , author Skakun, S. , author Shelestov, A. , author Lavreniuk, M. , author Yailymov, B. , author Kussul, O. , year 2015 . title Regional scale crop mapping using multi-temporal satellite imagery . journal The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences volume 40 , pages 45--52

    work page 2015

  45. [46]

    , author Tourre, Y

    author Lacaux, J. , author Tourre, Y. , author Vignolles, C. , author Ndione, J. , author Lafaye, M. , year 2007 . title Classification of ponds from high-spatial resolution remote sensing: Application to rift valley fever epidemics in senegal . journal Remote sensing of environment volume 106 , pages 66--74

    work page 2007

  46. [47]

    , author Wang, X

    author Lei, L. , author Wang, X. , author Zhang, L. , author Hu, X. , author Zhong, Y. , year 2024 . title Cropup: Historical products are all you need? an end-to-end cross-year crop map updating framework without the need for in situ samples . journal Remote Sensing of Environment volume 315 , pages 114430

    work page 2024

  47. [48]

    , author Abrams, J.F

    author Lenton, T.M. , author Abrams, J.F. , author Bartsch, A. , author Bathiany, S. , author Boulton, C.A. , author Buxton, J.E. , author Conversi, A. , author Cunliffe, A.M. , author Hebden, S. , author Lavergne, T. , et al., year 2024 . title Remotely sensing potential climate change tipping points across scales . journal nature communications volume 1...

    work page 2024

  48. [49]

    , author Di, L

    author Li, H. , author Di, L. , author Zhang, C. , author Lin, L. , author Guo, L. , author Yu, E.G. , author Yang, Z. , year 2024 a. title Automated in-season crop-type data layer mapping without ground truth for the conterminous united states based on multisource satellite imagery . journal IEEE Transactions on Geoscience and Remote Sensing volume 62 , ...

    work page doi:10.1109/tgrs.2024.3361895 2024

  49. [50]

    , author Chen, B

    author Li, Z. , author Chen, B. , author Wu, S. , author Su, M. , author Chen, J.M. , author Xu, B. , year 2024 b. title Deep learning for urban land use category classification: A review and experimental assessment . journal Remote Sensing of Environment volume 311 , pages 114290

    work page 2024

  50. [51]

    , author Kiefer, R.W

    author Lillesand, T. , author Kiefer, R.W. , author Chipman, J. , year 2015 . title Remote sensing and image interpretation . publisher John Wiley & Sons

    work page 2015

  51. [52]

    , author Zhong, R

    author Lin, Z. , author Zhong, R. , author Xiong, X. , author Guo, C. , author Xu, J. , author Zhu, Y. , author Xu, J. , author Ying, Y. , author Ting, K. , author Huang, J. , et al., year 2022 . title Large-scale rice mapping using multi-task spatiotemporal deep learning and sentinel-1 sar time series . journal Remote Sensing volume 14 , pages 699

    work page 2022

  52. [53]

    , author Sun, Y

    author Liu, C. , author Sun, Y. , author Xu, Y. , author Sun, Z. , author Zhang, X. , author Lei, L. , author Kuang, G. , year 2024 . title A review of optical and sar image deep feature fusion in semantic segmentation . journal IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

    work page 2024

  53. [54]

    , author Huete, A

    author Liu, H.Q. , author Huete, A. , year 1995 . title A feedback based modification of the ndvi to minimize canopy background and atmospheric noise . journal IEEE Transactions on Geoscience and Remote Sensing volume 33 , pages 457--465 . :10.1109/TGRS.1995.8746027

    work page doi:10.1109/tgrs.1995.8746027 1995

  54. [55]

    , author Zhang, L

    author Liu, Z. , author Zhang, L. , author Yu, Y. , author Xi, X. , author Ren, T. , author Zhao, Y. , author Zhu, D. , author Zhu, A.x. , year 2022 . title Cross-year reuse of historical samples for crop mapping based on environmental similarity . journal Frontiers in Plant Science volume 12 , pages 761148

    work page 2022

  55. [56]

    , author Chen, S

    author Ma, Y. , author Chen, S. , author Ermon, S. , author Lobell, D.B. , year 2024 . title Transfer learning in environmental remote sensing . journal Remote Sensing of Environment volume 301 , pages 113924

    work page 2024

  56. [57]

    , author Podra s c anin, Z

    author Mimi \'c , G. , author Podra s c anin, Z. , author Basarin, B. , year 2024 . title Change detection of the k \"o ppen climate zones in southeastern europe . journal Atmospheric Science Letters volume 25 , pages e1270

    work page 2024

  57. [58]

    , author Belgiu, M

    author Mohammadi, S. , author Belgiu, M. , author Stein, A. , year 2024 a. title Few-shot learning for crop mapping from satellite image time series . journal Remote Sensing volume 16 , pages 1026

    work page 2024

  58. [59]

    , author Belgiu, M

    author Mohammadi, S. , author Belgiu, M. , author Stein, A. , year 2024 b. title A source-free unsupervised domain adaptation method for cross-regional and cross-time crop mapping from satellite image time series . journal Remote Sensing of Environment volume 314 , pages 114385 . https://www.sciencedirect.com/science/article/pii/S0034425724004115, :https:...

    work page doi:10.1016/j.rse.2024.114385 2024

  59. [60]

    title Monthly synoptic weather report - december 2022

    author National Centers for Environmental Information , year 2022 . title Monthly synoptic weather report - december 2022 . https://www.ncei.noaa.gov/access/monitoring/monthly-report/synoptic/202213. note accessed: 2025-04-29

    work page 2022

  60. [61]

    , author Ho Tong Minh, D

    author Ndikumana, E. , author Ho Tong Minh, D. , author Baghdadi, N. , author Courault, D. , author Hossard, L. , year 2018 . title Deep recurrent neural network for agricultural classification using multitemporal sar sentinel-1 for camargue, france . journal Remote Sensing volume 10 , pages 1217 . https://www.mdpi.com/2072-4292/10/8/1217

    work page 2018

  61. [62]

    , author Mutavd z i \'c , B

    author Novkovi \'c , N. , author Mutavd z i \'c , B. , author Vukeli \'c , N. , year 2013 . title Vojvodina’s agriculture--analysis and possibilities

    work page 2013

  62. [63]

    , author Mrziglod, J

    author Nowakowski, A. , author Mrziglod, J. , author Spiller, D. , author Bonifacio, R. , author Ferrari, I. , author Mathieu, P.P. , author Garcia-Herranz, M. , author Kim, D.H. , year 2021 . title Crop type mapping by using transfer learning . journal International Journal of Applied Earth Observation and Geoinformation volume 98 , pages 102313 . https:...

    work page doi:10.1016/j.jag.2021.102313 2021

  63. [64]

    , author Pelletier, C

    author Nyborg, J. , author Pelletier, C. , author Lef \`e vre, S. , author Assent, I. , year 2022 . title Timematch: Unsupervised cross-region adaptation by temporal shift estimation . journal ISPRS Journal of Photogrammetry and Remote Sensing volume 188 , pages 301--313

    work page 2022

  64. [65]

    Papadakis, C., Filandrianos, G., Dimitriou, A., Lymperaiou, M., Thomas, K., Stamou, G., 2025

    author Pan, S.J. , author Yang, Q. , year 2010 . title A survey on transfer learning . journal IEEE Transactions on Knowledge and Data Engineering volume 22 , pages 1345--1359 . :10.1109/TKDE.2009.191

    work page doi:10.1109/tkde.2009.191 2010

  65. [66]

    , author Pavlović, D

    author Pandžić, M. , author Pavlović, D. , author Matavulj, P. , author Brdar, S. , author Marko, O. , author Crnojević, V. , author Kilibarda, M. , year 2024 . title Interseasonal transfer learning for crop mapping using sentinel-1 data . journal International Journal of Applied Earth Observation and Geoinformation volume 128 , pages 103718 . https://www...

    work page doi:10.1016/j.jag.2024.103718 2024

  66. [67]

    , author Turkoglu, M.O

    author Perich, G. , author Turkoglu, M.O. , author Graf, L.V. , author Wegner, J.D. , author Aasen, H. , author Walter, A. , author Liebisch, F. , year 2023 . title Pixel-based yield mapping and prediction from sentinel-2 using spectral indices and neural networks . journal Field Crops Research volume 292 , pages 108824 . https://www.sciencedirect.com/sci...

    work page doi:10.1016/j.fcr.2023.108824 2023

  67. [68]

    , author Inglada, J

    author Petitjean, F. , author Inglada, J. , author Gan c arski, P. , year 2012 . title Satellite image time series analysis under time warping . journal IEEE transactions on geoscience and remote sensing volume 50 , pages 3081--3095

    work page 2012

  68. [69]

    , author Ngugi, M.K

    author Peña-Barragán, J.M. , author Ngugi, M.K. , author Plant, R.E. , author Six, J. , year 2011 . title Object-based crop identification using multiple vegetation indices, textural features and crop phenology . journal Remote Sensing of Environment volume 115 , pages 1301--1316 . https://www.sciencedirect.com/science/article/pii/S0034425711000290, :http...

    work page doi:10.1016/j.rse.2011.01.009 2011

  69. [70]

    , author Tetteh, G

    author Pham, V.D. , author Tetteh, G. , author Thiel, F. , author Erasmi, S. , author Schwieder, M. , author Frantz, D. , author van der Linden, S. , year 2024 . title Temporally transferable crop mapping with temporal encoding and deep learning augmentations . journal International Journal of Applied Earth Observation and Geoinformation volume 129 , pages 103867

    work page 2024

  70. [71]

    , author Brdar, S

    author Radulovi \'c , M. , author Brdar, S. , author Pejak, B. , author Lugonja, P. , author Athanasiadis, I. , author Pajevi \'c , N. , author Pavi \'c , D. , author Crnojevi \'c , V. , year 2023 . title Machine learning-based detection of irrigation in vojvodina (serbia) using sentinel-2 data . journal GIScience & Remote Sensing volume 60 , pages 2262010

    work page 2023

  71. [72]

    , author Markovi \'c , M

    author Radulovi \'c , M. , author Markovi \'c , M. , author Brdar, S. , author Athanasiadis, I. , author Mimi \'c , G. , year 2025 a. title The first geospatial dataset of irrigated fields (2020--2024) in vojvodina (serbia) . journal Scientific Data volume 12 , pages 105

    work page 2025

  72. [73]

    , author Markovi \'c , M

    author Radulovi \'c , M. , author Markovi \'c , M. , author Brdar, S. , author Athanasiadis, I. , author Mimi \'c , G. , year 2025 b. title The first geospatial dataset of irrigated fields (2020--2024) in vojvodina (serbia) . journal Scientific Data volume 12 , pages 105

    work page 2025

  73. [74]

    , author Di, L

    author Rahman, M.S. , author Di, L. , author Yu, E. , author Zhang, C. , author Mohiuddin, H. , year 2019 . title In-season major crop-type identification for us cropland from landsat images using crop-rotation pattern and progressive data classification . journal Agriculture volume 9 , pages 17

    work page 2019

  74. [75]

    , author Ghosh, R

    author Ravirathinam, P. , author Ghosh, R. , author Khandelwal, A. , author Jia, X. , author Mulla, D. , author Kumar, V. , year 2024 . title Combining satellite and weather data for crop type mapping: An inverse modelling approach . journal arXiv [cs.CV] http://arxiv.org/abs/2401.15875

    work page arXiv 2024

  75. [76]

    , author Kasanick \`y , T

    author Rus n \'a k, T. , author Kasanick \`y , T. , author Mal \' k, P. , author Moj z i s , J. , author Zelenka, J. , author Svi c ek, M. , author Abrah \'a m, D. , author Halabuk, A. , year 2023 . title Crop mapping without labels: Investigating temporal and spatial transferability of crop classification models using a 5-year sentinel-2 series and machi...

    work page 2023

  76. [77]

    , author Wang, S

    author Ru wurm, M. , author Wang, S. , author Korner, M. , author Lobell, D. , year 2020 . title Meta-learning for few-shot land cover classification , in: booktitle Proceedings of the ieee/cvf conference on computer vision and pattern recognition workshops , pp. pages 200--201

    work page 2020

  77. [78]

    , author Körner, M

    author Rußwurm, M. , author Körner, M. , year 2020 . title Self-attention for raw optical satellite time series classification . journal ISPRS Journal of Photogrammetry and Remote Sensing volume 169 , pages 421--435 . https://www.sciencedirect.com/science/article/pii/S0924271620301647, :https://doi.org/10.1016/j.isprsjprs.2020.06.006

    work page doi:10.1016/j.isprsjprs.2020.06.006 2020

  78. [79]

    , author Ghazaryan, G

    author Safonova, A. , author Ghazaryan, G. , author Stiller, S. , author Main-Knorn, M. , author Nendel, C. , author Ryo, M. , year 2023 . title Ten deep learning techniques to address small data problems with remote sensing . journal International Journal of Applied Earth Observation and Geoinformation volume 125 , pages 103569

    work page 2023

  79. [80]

    , author Erasmi, S

    author Schlund, M. , author Erasmi, S. , year 2020 . title Sentinel-1 time series data for monitoring the phenology of winter wheat . journal Remote Sensing of Environment volume 246 , pages 111814

    work page 2020

  80. [81]

    , author Paliwal, K.K

    author Schuster, M. , author Paliwal, K.K. , year 1997 . title Bidirectional recurrent neural networks . journal IEEE transactions on Signal Processing volume 45 , pages 2673--2681

    work page 1997

Showing first 80 references.