Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

REO-VLM: Transforming VLM to Meet Regression Challenges in Earth Observation

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 vision-language model can both describe satellite images and estimate above-ground biomass from them.

desk verdict A large and honestly reported EO benchmark, but the 'knowledge-driven' regression claim is confounded by likely label leakage and asymmetric baselines. read the letter →

arxiv 2412.16583 v1 pith:QRU5OQVM submitted 2024-12-21 cs.CV

Xizhe Xue , Guoting Wei , Hao Chen , Haokui Zhang , Feng Lin , Chunhua Shen , Xiao Xiang Zhu This is my paper · ORCID
classification cs.CV
keywords vision-languagemodelsEarthobservationabove-groundbiomassregressionmultimodalremotesensinginstructiontuningbenchmarkreverseprojectionmodulescientificSARandmultispectralfusion
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

This paper attempts to show that a vision-language model (VLM) can do more than describe satellite imagery: it can also perform scientific regression, namely estimating above-ground biomass (AGB) from multispectral and radar data. To that end it introduces REO-Instruct, a 1.6-million-pair benchmark that pairs RGB, multispectral, and SAR images with land-cover labels, ecological patch counts, human-activity question-answer pairs, and ground-truth AGB values. Building on this dataset, it proposes REO-VLM, a single model with separate generation and regression heads, trained in two stages so the language side first learns domain knowledge and the regression side then consumes it. The key reported result is a positive R-squared of 0.36 for AGB regression on multispectral input, where a fine-tuned LLaVA baseline is negative, which the authors read as evidence that the model captures real numerical patterns rather than memorizing answers. A sympathetic reader would care because it points toward VLMs as joint tools for environmental monitoring and resource management, not just captioning.

What carries the argument

The carrying mechanism is the reverse projection module (R-Proj), a linear layer that projects 4096-dimensional LLM hidden features back into the 1024-dimensional visual feature space, feeding language-derived contextual information into a four-layer MLP-mixer-style regression head. Around it sit three supporting components: spectral recombination, which recasts multispectral bands and SAR polarization channels as pseudo-RGB images so the frozen visual encoder can consume them; visual token selection, which harvests features from middle layers of the visual encoder so regression gets fine detail rather than only high-level semantic tokens; and a two-stage training scheme that first fine-tunes the LLM and generation head with cross-entropy, then trains only the regression head and reverse projection module with MSE loss, decoupling the conflicting objectives of discrete generation and continuous regression.

What would settle it

Train REO-VLM on a version of REO-Instruct in which all numeric AGB values are removed or paraphrased out of the text annotations, keeping land-cover and human-activity text intact, then measure AGB R-squared on the same test set. If R-squared stays at roughly 0.36, the reverse projection is genuinely transferring domain knowledge; if it collapses toward the fine-tuned LLaVA baseline's negative value, the result depended on label leakage through the language pathway.

Watch

Extended reading notes

Core claim

On the authors' own terms, REO-VLM is the first unified Earth-observation VLM that integrates scientific regression and generative capabilities in a single model. Using REO-Instruct, the model is trained first with the LLM and generation head, then with a regression head and a reverse projection module that maps LLM hidden states back into visual feature space. In the authors' experiments, this design yields an R-squared of 0.20 (RGB), 0.36 (MS), and 0.35 (MS+SAR) for AGB regression, outperforming a fine-tuned LLaVA baseline whose R-squared is negative across modalities, while also handling land-cover classification, patch counting, and human-activity VQA. The paper claims these positive R-squared values prove the model captures underlying numerical patterns in the data, and that the language-driven reasoning supplies scientific domain knowledge that breaks the information bottleneck of image-only regression.

Load-bearing premise

The claimed advantage of the reverse projection module rests on the assumption that the LLM's hidden states carry scientific knowledge that is independent of the numeric answer being predicted, rather than a copy of the ground-truth AGB value that appears in the training annotations.

Editorial extensions

If this is right

  • If the central claim holds, a single VLM can jointly answer questions, classify land cover, count ecological patches, and estimate biomass from the same satellite scenes, simplifying EO analysis pipelines.
  • Positive R-squared on AGB regression with multispectral input implies VLMs are a viable alternative to dedicated image-only regression models, narrowing the gap with specialized baselines like Niconet while adding descriptive abilities.
  • The two-stage decoupling strategy suggests that regression and generation objectives need not be reconciled within one loss; separating them by training phase may transfer to other scientific regression tasks in EO, such as species range or population density estimation.
  • The REO-Instruct benchmark itself provides a reusable training and evaluation substrate for future EO-VLMs aiming at regression-plus-generation.
  • The reverse projection principle, pulling LLM-generated knowledge back into the visual space, could be adapted to other multimodal regression problems beyond remote sensing.

Reading between the lines

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

  • The paper's own setup leaves open a label-leakage path: the REO-Instruct text annotations contain the ground-truth AGB values, and stage one fine-tunes the LLM on those annotations. If the LLM hidden states fed to the regression head already encode the numeric answer being generated, the reported gain over the LLaVA baseline could partly come from copying the label through the reverse projection,
  • Because the authors report negative R-squared for ecological patch counting when numbers are treated as text tokens, a natural extension of their thesis is that any VLM regression pipeline should route numeric outputs through a dedicated regression head, and that token-level generation alone is insufficient for continuous scientific quantities.
  • The reliance on a commercial language model to generate text annotations is a potential reproducibility vulnerability: the benchmark's scientific content depends on the exact prompt templates and model version used, and a systematic release of those prompts would make the dataset's construction fully auditable.
  • If the mechanism generalizes, the same architecture could be applied to other continuous geophysical attributes such as soil moisture, canopy height, or surface temperature, with the reverse projection providing the knowledge bridge.
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. The paper introduces REO-Instruct, a large Earth-Observation benchmark with 1.6 million multimodal (RGB, multispectral, SAR) image-text pairs spanning four tasks: land cover classification, ecological patch counting, VQA-based human activity monitoring, and above-ground biomass (AGB) regression. It then proposes REO-VLM, a LLaVA-1.5-based architecture with a dedicated regression head, a reverse projection module that maps LLM hidden states back into the visual feature space, and a two-stage training strategy: stage one fine-tunes the LLM on text annotations with cross-entropy loss, and stage two trains the regression head and reverse projection with MSE loss. On the REO-Instruct test split, the paper reports R²=0.36 for AGB regression with multispectral input, 19.94% overall accuracy on land cover classification, 80.50% on VQA, and negative R² for patch counting. The central claim is that the reverse projection mechanism lets the model exploit scientific domain knowledge stored in the LLM to improve regression accuracy.

Significance. If validated, the paper would make a useful contribution: REO-Instruct appears to be one of the largest EO instruction datasets that explicitly pairs regression targets with rich text annotations, and the two-stage separation of generation and regression training is a sensible way to address conflicting optimization objectives. The paper is also candid about the patch-counting failure, which is a point in its favor. However, the central scientific claim—that language-driven scientific knowledge, rather than memorization of the regression target, is what improves AGB prediction—is not established by the reported experiments. The comparison protocol is asymmetric, the absolute regression accuracy is modest (best R²=0.36), and one of the two regression tasks in the benchmark has negative R² for every model including REO-VLM. With additional controls and a fair baseline protocol, the contribution could be significant; as it stands, the evidence is not sufficient to support the headline claims.

major comments (4)
  1. [Tables 2–4 and Figure 5] The comparison protocol is asymmetric. REO-VLM is trained on the REO-Instruct training split in both stages, while GeoChat, LHRS-Bot, Qwen2-VL, and ChatGPT-4o are evaluated zero-shot with only guiding prompts. The only baseline fine-tuned on REO-Instruct is LLaVA, and it appears only in the AGB regression table. The large gaps (e.g., VQA accuracy 80.50% vs. 33.79% for ChatGPT-4o; land-cover OA 19.94% vs. 3.97%) therefore do not establish that REO-VLM outperforms these systems; the baselines have not been given the opportunity to adapt to the task distribution. The paper should either fine-tune all baselines on the same training split or explicitly restrict all comparative claims to the fine-tuned setting.
  2. [§3.4, §4.1, §4.2] The design does not rule out label leakage through the reverse projection. Section 3.4 states that the text annotations contain "quantitative ground-truth estimates of the Above-Ground Biomass," and Section 4.2 says stage one fine-tunes the LLM on these annotations with cross-entropy loss. The reverse projection module in Section 4.1 then reads 4096-dimensional LLM hidden states and projects them into the visual feature space consumed by the regression head. If those hidden states already encode the AGB value that the LLM was trained to generate, the regression head may be reading a memorized label out of the language model rather than integrating scientific domain knowledge. This is load-bearing because the knowledge-driven mechanism is the paper's stated novelty. The paper needs at least three controls: (i) an ablation without the reverse projection; (ii) a training run in which AGB values are withheld from or masked in the text annotations during stage one; and (iii) a test of whether the regression head can recover the target from hidden states of a frozen or randomly initialized LLM.
  3. [§5.3 and Table 3] The ecological patch counting task fails for every model, including REO-VLM, with negative R² in all rows of Table 3. Since patch counting is one of the four core tasks in REO-Instruct and is listed as a supported capability in Table 1, this is a failure on one of the benchmark's own tasks, not a minor limitation. The paper's explanation that numeric values were "treated as texts" and trained with cross-entropy does not apply to REO-VLM, which has a dedicated regression head and MSE loss in stage two; the authors should explain why the regression head was not used for counting, and they should not claim that REO-VLM unifies regression and generation while one of the two regression tasks has negative R².
  4. [§5.4 and Table 4] The AGB regression results are not compared against the full set of baselines. Table 4 includes only LLaVA variants for the direct comparison, yet the text states that "almost all comparison algorithms fail to perform effective AGB regression"—GeoChat, LHRS-Bot, Qwen2-VL, and ChatGPT-4o are not evaluated on this task. Moreover, the authors cite Niconet's RMSE of 69.0 on the same dataset, whereas REO-VLM's best RMSE is 75.59; the paper therefore does not demonstrate state-of-the-art regression accuracy. The R²=0.36 result should be reported as a weak positive correlation, not as proof that the model "captures underlying numerical patterns," especially given the label-leakage confound described above.
minor comments (5)
  1. [§4.1 heading] The heading contains a typo: "Architectue" should be "Architecture."
  2. [Table 1] Table 1 marks the counting capability ("Cnt.") as supported for REO-VLM, but Table 3 reports negative R² for this task; the table should be qualified to avoid overstating the model's capabilities.
  3. [Table 5] The strategies "Half layers" and "Half layers (deep)" are not sufficiently distinguished in the text; the reader cannot tell exactly which layers are included in each configuration.
  4. [§3.4 and Figure 3] The manuscript does not describe how ChatGPT-4o outputs were checked for factual consistency with the land cover and AGB labels; a brief verification protocol would strengthen the benchmark's credibility.
  5. [§3.3 and §5] The paper states that the test set has about 36K pairs and then says each downstream test subset contains approximately 8.6K unique samples; it should clarify whether the 36K is the union of the four subsets and confirm that the four subsets are mutually exclusive.

Circularity Check

2 steps flagged · score 6.0 of 10

REO-VLM's claimed knowledge-driven AGB regression is confounded by construction: the LLM hidden states read by the reverse projection are trained on text containing the ground-truth AGB values, so the reported R^2 gain may be label re-readout rather than scientific reasoning.

  1. fitted input called prediction [Section 3.4; Section 4.1; Section 4.2]
    "The text annotations also provide quantitative ground-truth estimates of the Above-Ground Biomass (AGB) in the corresponding image areas. ... In the first stage, the LLM is trained along with the generation head. ... we fine-tune the LLM component and generation head of the pre-trained LLaVA-1.5 model using LoRA during this phase, with cross-entropy training loss. ... our module retrieves contextual information generated by the LLM and maps it back into the visual feature space to assist the regression process."

    Stage one trains the LLM to emit the exact AGB value that appears in the REO-Instruct text annotations. Stage two attaches a regression head to the LLM's hidden states via the reverse projection and trains it on the same AGB values. The reported R^2 improvement over fine-tuned LLaVA can therefore be obtained by decoding the already-memorized numeric label from the hidden states, rather than by integrating scientific domain knowledge. No ablation withholds AGB from the text or removes the reverse projection, so the central mechanism is confounded by construction.

  2. other [Section 5.4]
    "REO-VLM achieves a positive R-squared value (0.20 for RGB, 0.36 for MS, and 0.35 for MS+SAR), proving that the model captures underlying numerical patterns in the data."

    This 'proof' relies on the same confounded setup: because the LLM hidden states were fine-tuned on text containing ground-truth AGB estimates, a positive R^2 does not demonstrate discovery of numerical patterns; it may only demonstrate successful readout of the target from the LLM's next-token representations. The claim overstates what the experiment can establish without a control that removes the target from the language supervision.

full rationale

REO-VLM's central novelty is that language-driven reasoning injects scientific domain knowledge into regression, with the evidence being the R^2 improvement over fine-tuned LLaVA in Table 4. However, the construction permits a simpler explanation. REO-Instruct text annotations include the ground-truth AGB numbers, and stage one fine-tunes the LLM with cross-entropy on those annotations; the reverse projection then reads the LLM's hidden states and feeds them to the regression head, which is trained with MSE on the same AGB target. Because those hidden states were optimized to generate the exact target values as tokens, the regression head can recover the target from the LLM's next-token representations rather than from EO imagery plus independent scientific knowledge. The paper provides no ablation removing AGB from the text, no experiment withholding the target from stage one, and no analysis showing that hidden-state content beyond a label copy matters. Thus the central 'knowledge-driven' claim is partially circular: the prediction can reduce to a re-readout of labels already present in the training signal. This is not a formal mathematical circularity, and there is no load-bearing self-citation, but the main mechanistic claim is confounded by construction, warranting a score of 6.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central method rests on two hand-chosen design elements (visual token selection and spectral recombination), reliance on the AGBD labels, and on ChatGPT-4o text derived from the same labels. No new physical entities are introduced; the reverse projection is an architectural component, not a postulated entity.

free parameters (2)
  • Visual token selection strategy = Half layers (balanced shallow/deep)
    Selected based on ablation in Table 5 directly on the test set; no dedicated validation split is reported, so this is a post-hoc design choice that affects reported AGB performance.
  • Spectral recombination band combinations = [B05,B06,B07], [B12,B11,B02], [B04,B03,B02], [B08,B04,B03], [B12,B08,B04]
    Hand-chosen standard band triples; presented as a strategy to reuse the pretrained encoder, but no comparison against alternative band groupings is given.
assumptions (4)
  • domain assumption The AGBD dataset's AGB ground truth and Copernicus land cover labels are accurate enough to serve as supervision.
    All training and evaluation depend on these labels (Sections 3.3 and 3.4); the paper does not validate label quality beyond citing AGBD.
  • ad hoc to paper Text annotations generated by ChatGPT-4o from the same land cover and AGB data faithfully represent scientific knowledge and do not encode the regression target in a way that leaks through the reverse projection.
    The claimed benefit of the reverse projection module (Section 4.1) presupposes that the LLM hidden states convey independent domain knowledge; the paper provides no analysis separating leakage of the target label from genuine auxiliary information.
  • domain assumption The pseudo-RGB and spectral recombination conversions preserve the information needed for AGB regression.
    The visual encoder is frozen and pretrained on natural RGB images (Section 4.1); the paper assumes 25x25 patches and recombined bands are sufficient, without ablating against higher resolution or native multi-band encoders.
  • domain assumption LLaVA-1.5 pretrained weights are a suitable foundation for EO imagery.
    The architecture follows LLaVA-1.5 (Section 4.1) and relies on its pretrained initialization; no comparison with other foundation models is reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of REO-VLM: Transforming VLM to Meet Regression Challenges in Earth Observation." pith.science (2026). https://pith.science/paper/QRU5OQVM

@misc{pith2026241216583,
  author       = {Pith},
  title        = {Pith review of: REO-VLM: Transforming VLM to Meet Regression Challenges in Earth Observation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QRU5OQVM}},
  note         = {Machine review of arXiv:2412.16583}
}
read the original abstract

The rapid evolution of Vision Language Models (VLMs) has catalyzed significant advancements in artificial intelligence, expanding research across various disciplines, including Earth Observation (EO). While VLMs have enhanced image understanding and data processing within EO, their applications have predominantly focused on image content description. This limited focus overlooks their potential in geographic and scientific regression tasks, which are essential for diverse EO applications. To bridge this gap, this paper introduces a novel benchmark dataset, called \textbf{REO-Instruct} to unify regression and generation tasks specifically for the EO domain. Comprising 1.6 million multimodal EO imagery and language pairs, this dataset is designed to support both biomass regression and image content interpretation tasks. Leveraging this dataset, we develop \textbf{REO-VLM}, a groundbreaking model that seamlessly integrates regression capabilities with traditional generative functions. By utilizing language-driven reasoning to incorporate scientific domain knowledge, REO-VLM goes beyond solely relying on EO imagery, enabling comprehensive interpretation of complex scientific attributes from EO data. This approach establishes new performance benchmarks and significantly enhances the capabilities of environmental monitoring and resource management.

Figures

Figures reproduced from arXiv: 2412.16583 by the authors.

Figure 1
Figure 1. Motivations of VLMs for EO regression. (a). Hierarchi￾cal structure of VLM capabilities: From basic perception tasks to higher-order reasoning tasks; (b). Advantages of VLM for EO regression tasks: By integrating scientific domain knowledge with EO image data, VLMs overcome the information bottleneck of traditional image-only regression models, enabling deeper in￾sights and improved scientific reasoning; (c). Interp… view at source ↗
Figure 2
Figure 2. Image examples and prompt suite statistics of REO-Instruct benchmark. (a). Some image screenshots in RGB modality; (b). [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Screenshots of some image-texts annotation pairs in REO-Instruct benchmark. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Overall framework of proposed REO-VLM. G head and R head denote generation and regression heads respectively. R-Proj is [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Comparative experimental results on the VQA-human [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Qualitative experimental results of REO-VLM and other methods on different downstream tasks in REO-Instruct benchmark. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Land cover distribution based on the number of samples [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]
Figure 8
Figure 8. Figure 8: Distribution of Above-Ground Biomass (AGB) values in [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: Test guiding prompts for compared methods. [PITH_FULL_IMAGE:figures/full_fig_p014_9.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. OSMDA: OpenStreetMap-based Domain Adaptation for Remote Sensing VLMs

    cs.CV 2026-03 conditional novelty 6.0 of 10

    A remote-sensing VLM can be adapted by having it read rendered OpenStreetMap maps paired with satellite images, then fine-tuning it on satellite images alone.

Reference graph

Works this paper leans on

37 extracted references · 22 canonical work pages · cited by 1 Pith paper

  1. [1]

    Gpt-4 technical report

    Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ah- mad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774 ,

    arXiv

  2. [2]

    M3leo: A multi-modal, multi-label earth observation dataset integrating interfero- metric sar and multispectral data

    Matthew J Allen, Francisco Dorr, Joseph Alejandro Gal- lego Mejia, Laura Mart´ınez-Ferrer, Anna Jungbluth, Freddie Kalaitzis, and Ra ´ul Ramos-Poll´an. M3leo: A multi-modal, multi-label earth observation dataset integrating interfero- metric sar and multispectral data. In The Thirty-eight Con- ference on Neural Information Processing Systems Datasets and ...

    work page 2024

  3. [3]

    Uncovering temporal changes in europe’s population density patterns using a data fusion approach

    Filipe Batista e Silva, S ´ergio Freire, Marcello Schiavina, Konˇstant´ın Rosina, Mario Alberto Mar ´ın-Herrera, Lukasz Ziemba, Massimo Craglia, Eric Koomen, and Carlo Lavalle. Uncovering temporal changes in europe’s population density patterns using a data fusion approach. Nature communica- tions, 11(1):4631, 2020. 1

    work page 2020

  4. [4]

    Rs-llava: A large vision-language model for joint captioning and question an- swering in remote sensing imagery

    Yakoub Bazi, Laila Bashmal, Mohamad Mahmoud Al Rah- hal, Riccardo Ricci, and Farid Melgani. Rs-llava: A large vision-language model for joint captioning and question an- swering in remote sensing imagery. Remote Sensing, 16(9): 1477, 2024. 2

    work page 2024

  5. [5]

    Multi-modal learning for geospatial vegetation forecasting

    Vitus Benson, Claire Robin, Christian Requena-Mesa, Lazaro Alonso, Nuno Carvalhais, Jos ´e Cort´es, Zhihan Gao, Nora Linscheid, M´elanie Weynants, and Markus Reichstein. Multi-modal learning for geospatial vegetation forecasting. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 27788–27799, 2024. 1, 2, 3

    work page 2024

  6. [6]

    Combining observational data and language for species range estimation

    Max Hamilton, Christian Lange, Elijah Cole, Alexan- der Shepard, Samuel Heinrich, Oisin Mac Aodha, Grant Van Horn, and Subhransu Maji. Combining observational data and language for species range estimation. NIPS, 2024. 1, 2, 3

    work page 2024

  7. [7]

    Rsgpt: A remote sensing vision language model and benchmark

    Yuan Hu, Jianlong Yuan, Congcong Wen, Xiaonan Lu, and Xiang Li. Rsgpt: A remote sensing vision language model and benchmark. arXiv preprint arXiv:2307.15266, 2023. 2, 3

    arXiv 2023

  8. [8]

    Geochat: Grounded large vision-language model for remote sensing

    Kartik Kuckreja, Muhammad Sohail Danish, Muzammal Naseer, Abhijit Das, Salman Khan, and Fahad Shahbaz Khan. Geochat: Grounded large vision-language model for remote sensing. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 27831– 27840, 2024. 2, 3, 6, 8

    work page 2024

Show all 37 references
  1. [9]

    A high-resolution canopy height model of the earth

    Nico Lang, Walter Jetz, Konrad Schindler, and Jan Dirk Wegner. A high-resolution canopy height model of the earth. Nature Ecology & Evolution, 7(11):1778–1789, 2023. 1, 2, 3, 10

    2023

  2. [10]

    Above-ground biomass prediction by sentinel-1 multitemporal data in cen- tral italy with integration of alos2 and sentinel-2 data

    Gaia Vaglio Laurin, Johannes Balling, Piermaria Corona, Walter Mattioli, Dario Papale, Nicola Puletti, Maria Rizzo, John Truckenbrodt, and Marcel Urban. Above-ground biomass prediction by sentinel-1 multitemporal data in cen- tral italy with integration of alos2 and sentinel-2...

  3. [11]

    Forest aboveground biomass estimation using landsat 8 and sentinel-1a data with machine learning algorithms

    Yingchang Li, Mingyang Li, Chao Li, and Zhenzhen Liu. Forest aboveground biomass estimation using landsat 8 and sentinel-1a data with machine learning algorithms. Scientific reports, 10(1):9952, 2020. 3, 4

    2020

  4. [12]

    Re- moteclip: 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. Re- moteclip: A vision language foundation model for remote sensing. IEEE Transactions on Geoscience and Remote Sensing, 2024. 2, 3

    2024

  5. [13]

    Visual instruction tuning

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. Visual instruction tuning. Advances in neural information processing systems, 36, 2024. 1, 8

    2024

  6. [14]

    Skysensegpt: A fine-grained in- struction tuning dataset and model for remote sensing vision- language understanding

    Junwei Luo, Zhen Pang, Yongjun Zhang, Tingzhu Wang, Linlin Wang, Bo Dang, Jiangwei Lao, Jian Wang, Jingdong Chen, Yihua Tan, et al. Skysensegpt: A fine-grained in- struction tuning dataset and model for remote sensing vision- language understanding. arXiv preprint arXiv:2406.10100,

    arXiv

  7. [15]

    The global distri- bution and environmental drivers of aboveground versus be- 11 lowground plant biomass

    Haozhi Ma, Lidong Mo, Thomas W Crowther, Daniel S Maynard, Johan van den Hoogen, Benjamin D Stocker, C´esar Terrer, and Constantin M Zohner. The global distri- bution and environmental drivers of aboveground versus be- 11 lowground plant biomass. Nature Ecology & Evolution , 5...

    2021

  8. [16]

    Re- mote sensing vision-language foundation models without annotations via ground remote alignment

    Utkarsh Mall, Cheng Perng Phoo, Meilin Kelsey Liu, Carl V ondrick, Bharath Hariharan, and Kavita Bala. Re- mote sensing vision-language foundation models without annotations via ground remote alignment. arXiv preprint arXiv:2312.06960, 2023. 2, 3

    2023 arXiv

  9. [17]

    Lhrs-bot: Empowering remote sensing with vgi-enhanced large multimodal language model

    Dilxat Muhtar, Zhenshi Li, Feng Gu, Xueliang Zhang, and Pengfeng Xiao. Lhrs-bot: Empowering remote sensing with vgi-enhanced large multimodal language model. arXiv preprint arXiv:2402.02544, 2024. 2, 3, 8

    2024 arXiv

  10. [18]

    Climatelearn: Benchmarking machine learning for weather and climate modeling

    Tung Nguyen, Jason Jewik, Hritik Bansal, Prakhar Sharma, and Aditya Grover. Climatelearn: Benchmarking machine learning for weather and climate modeling. Advances in Neural Information Processing Systems, 36, 2024. 1

    2024

  11. [19]

    Training language models to follow instructions with human feedback

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Car- roll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. Training language models to follow instructions with human feedback. Ad- vances in neural information processing systems, 35:...

    2022

  12. [20]

    H2rsvlm: Towards helpful and honest re- mote sensing large vision language model

    Chao Pang, Jiang Wu, Jiayu Li, Yi Liu, Jiaxing Sun, Wei- jia Li, Xingxing Weng, Shuai Wang, Litong Feng, Gui- Song Xia, et al. H2rsvlm: Towards helpful and honest re- mote sensing large vision language model. arXiv preprint arXiv:2403.20213, 2024. 2

    2024 arXiv

  13. [21]

    Learning transferable visual models from natural language supervi- sion

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervi- sion. In International conference on machine learning, ...

    2021

  14. [22]

    Lidar-based reference aboveground biomass maps for tropical forests of south asia and central africa.Sci- entific Data, 11(1):334, 2024

    Suraj Reddy Rodda, Rakesh Fararoda, Rajashekar Gopalakr- ishnan, Nidhi Jha, Maxime R´ejou-M´echain, Pierre Couteron, Nicolas Barbier, Alonso Alfonso, Ousmane Bako, Patrick Bassama, et al. Lidar-based reference aboveground biomass maps for tropical forests of south asia and cen...

    2024

  15. [23]

    Agbd: A global-scale biomass dataset

    Ghjulia Sialelli, Torben Peters, Jan D Wegner, and Konrad Schindler. Agbd: A global-scale biomass dataset. arXiv preprint arXiv:2406.04928, 2024. 1, 3, 4, 13

    2024 arXiv

  16. [24]

    Large language models for captioning and retrieving remote sensing images

    Jo ˜ao Daniel Silva, Jo ˜ao Magalh ˜aes, Devis Tuia, and Bruno Martins. Large language models for captioning and retrieving remote sensing images. arXiv preprint arXiv:2402.06475, 2024. 2

    2024 arXiv

  17. [25]

    Mlp-mixer: An all-mlp architecture for vision

    Ilya O Tolstikhin, Neil Houlsby, Alexander Kolesnikov, Lu- cas Beyer, Xiaohua Zhai, Thomas Unterthiner, Jessica Yung, Andreas Steiner, Daniel Keysers, Jakob Uszkoreit, et al. Mlp-mixer: An all-mlp architecture for vision. Advances in neural information processing systems, 34:2...

  18. [26]

    Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution

    Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, et al. Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution. arXiv preprint arXiv:2409.12191, 2024. 8

    2024 arXiv

  19. [27]

    Skyscript: A large and seman- tically diverse vision-language dataset for remote sensing

    Zhecheng Wang, Rajanie Prabha, Tianyuan Huang, Jiajun Wu, and Ram Rajagopal. Skyscript: A large and seman- tically diverse vision-language dataset for remote sensing. In Proceedings of the AAAI Conference on Artificial Intel- ligence, pages 5805–5813, 2024. 2, 3

    2024

  20. [28]

    Torchspatial: A location encoding framework and benchmark for spatial representation learning

    Nemin Wu, Qian Cao, Zhangyu Wang, Zeping Liu, Yanlin Qi, Jielu Zhang, Joshua Ni, Xiaobai Yao, Hongxu Ma, Lan Mu, et al. Torchspatial: A location encoding framework and benchmark for spatial representation learning. NIPS, 2024. 2, 3

    2024

  21. [29]

    Skyeyegpt: Unifying remote sensing vision-language tasks via instruc- tion tuning with large language model

    Yang Zhan, Zhitong Xiong, and Yuan Yuan. Skyeyegpt: Unifying remote sensing vision-language tasks via instruc- tion tuning with large language model. arXiv preprint arXiv:2401.09712, 2024. 2, 3

    2024 arXiv

  22. [30]

    Earthgpt: A universal multi-modal large lan- guage model for multi-sensor image comprehension in re- mote sensing domain

    Wei Zhang, Miaoxin Cai, Tong Zhang, Yin Zhuang, and Xuerui Mao. Earthgpt: A universal multi-modal large lan- guage model for multi-sensor image comprehension in re- mote sensing domain. IEEE Transactions on Geoscience and Remote Sensing, 2024. 2, 3, 6

    2024

  23. [31]

    Rs5m and georsclip: A large scale vision-language dataset and a large vision-language model for remote sensing

    Zilun Zhang, Tiancheng Zhao, Yulong Guo, and Jianwei Yin. Rs5m and georsclip: A large scale vision-language dataset and a large vision-language model for remote sensing. IEEE Transactions on Geoscience and Remote Sensing, 2024. 3

    2024

  24. [32]

    Towards vision- language geo-foundation model: A survey

    Yue Zhou, Litong Feng, Yiping Ke, Xue Jiang, Junchi Yan, Xue Yang, and Wayne Zhang. Towards vision- language geo-foundation model: A survey. arXiv preprint arXiv:2406.09385, 2024. 2 12 REO-VLM: Transforming VLM to Meet Regression Challenges in Earth Observation Supplementary M...

    2024

  25. [33]

    The categories, ranging from ”Cultivated and managed veg- etation” to ”Bare sparse vegetation,” are sorted in descend- ing order

    Land Cover Distribution in REO-Instruct Figure 7 illustrates the distribution of land cover categories based on the number of samples collected for each type. The categories, ranging from ”Cultivated and managed veg- etation” to ”Bare sparse vegetation,” are sorted in descend-...

  26. [34]

    The data distributions in the two sets are highly similar, ensuring consistency and fairness be- tween the two sets

    Above Ground Biomass Value Distribution Figure 8 shows the distribution of AGB values in both the training and test sets. The data distributions in the two sets are highly similar, ensuring consistency and fairness be- tween the two sets. The AGB values range from [0-500], fol...

  27. [35]

    By selecting appropriate band combinations, we Figure 8

    Spectral Recombination Strategy In the field of EO, band combinations are commonly used to better interpret image features by enhancing specific char- acteristics. By selecting appropriate band combinations, we Figure 8. Distribution of Above-Ground Biomass (AGB) values in REO...

  28. [36]

    Every layer in the regression head includes two main stages: 4.1

    Regression Head Architecture The visual tokens from the encoder and the hidden tokens from the LLM are combined and processed through the re- gression head. Every layer in the regression head includes two main stages: 4.1. Token Aggregation Stage This stage fuses intra-token f...

  29. [37]

    These prompts shown in Figure 9 assist VLMs in better under- standing and addressing multiple tasks

    Guiding Prompts To define the scope of our questions and expected answers, we provide guiding prompts during testing for all compar- ison algorithms, except for our proposed method. These prompts shown in Figure 9 assist VLMs in better under- standing and addressing multiple tasks. 14

Pith tools

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