LongEarth-R1: Benchmarking and Aligning Vision-Language Models for Long-Horizon Earth Observation Reasoning

Chaofeng Chen; Gui-Song Xia; Jing Xiao; Liang Liao; Mi Wang; Yupan Ding; Zhenyuan Zhang

REVIEW 3 major objections 4 minor 47 references

LongEarth-R1: Benchmarking and Aligning Vision-Language Models for Long-Horizon Earth Observation Reasoning

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a vision-language model trained on long satellite image sequences with temporal and spatial rewards—LongEarth-R1—achieves the best results on all 12 long-horizon Earth observation reasoning tasks while staying…

desk verdict Useful benchmark, but the headline results are unverifiable because the model is trained and evaluated on the same data without a held-out split. read the letter →

arxiv 2608.13344 v1 pith:WL57E2XR submitted 2026-08-13 cs.AI

Yupan Ding , Jing Xiao , Zhenyuan Zhang , Chaofeng Chen , Liang Liao , Gui-Song Xia , Mi Wang This is my paper · ORCID

classification cs.AI

keywords long-horizonreasoningEarthobservationvision-languagemodelsremotesensingbenchmarkchain-of-thoughtgrouprelativepolicyoptimizationtemporalgroundingspatial

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

Long-horizon Earth observation reasoning means making sense of how a place changes across many satellite images over time, including stages of evolution, where and when changes happen, anomalies in the sequence, and what comes next. The paper argues that existing remote sensing vision-language models mostly handle single images, image pairs, or short clips, so they cannot reliably ground conclusions in the right frames and regions. It introduces LongEarth-Bench, about 120k question-answer samples built from 117k images, with sequences averaging 15.14 frames and reaching 30 frames, covering 12 tasks across four cognitive dimensions. It then trains LongEarth through supervised fine-tuning with explicit sequence identifiers and structured reasoning traces, and LongEarth-R1 through reinforcement learning with format, temporal, and spatial rewards. The central claim is that LongEarth-R1 achieves the best result on every one of the 12 long-sequence tasks while remaining competitive on standard remote sensing benchmarks.

What carries the argument

The load-bearing mechanism is the combination of explicit sequence identifiers and spatiotemporal rewards. Each frame is labeled Image 1 through Image T in the multimodal input, giving the model stable temporal anchors; a 30k-sample structured reasoning subset supplies chain-of-thought traces that connect key frames, changed regions, and final answers. On top of supervised fine-tuning, GRPO optimizes a weighted reward $R_i = \lambda_f R^{\mathrm{fmt}}_i + \lambda_t R^{\mathrm{time}}_i + \lambda_s R^{\mathrm{space}}_i$ that checks answer format, temporal ordering and key-frame matching, and spatial agreement on regions, directions, and extents. This machinery is what shifts the model from imitating answers to producing temporally ordered, spatially grounded reasoning.

What would settle it

Re-run the evaluation on a human-verified held-out subset of sequences whose labels are independently re-derived from the raw segmentation, polygon, land-cover, and flood evidence, and check whether LongEarth-R1 still leads on all 12 tasks; if a nontrivial fraction of the original labels contain wrong frame ranges, mislocated anomalies, or inconsistent spatial extents, the reported gains could be partly an artifact of the rule-based annotation pipeline.

Watch

Extended reading notes

Core claim

LongEarth-R1 establishes, on the paper's own evaluation, that long-horizon spatiotemporal reasoning in remote sensing can be trained and measured. On LongEarth-Bench it outperforms all compared baselines on all 12 tasks, with particularly strong gains on anomaly identification and tasks requiring long-range temporal evidence. LongEarth-R1 also transfers to conventional benchmarks, achieving the best result on six of nine single-image, bi-temporal, and short-sequence datasets, including all bi-temporal change-understanding benchmarks. The discovery is two-part: a benchmark that operationalizes long-horizon reasoning through four cognitive dimensions, and a training recipe—explicit sequence identifiers plus structured chain-of-thought, then group relative policy optimization with format, temporal, and spatial rewards—that turns a general vision-language model into a state-of-the-art long-sequence Earth observation reasoner.

Load-bearing premise

The benchmark's ground-truth answers and temporal and spatial labels, generated by rule-based pipelines from segmentation, polygon, land-cover, and flood evidence, are correct and unambiguous across all 12 tasks, so scores measured on it faithfully reflect long-horizon reasoning ability.

Editorial extensions

If this is right

Long-horizon Earth observation reasoning can be benchmarked and trained end-to-end, with sequences up to 30 frames tractable for current vision-language models when explicit temporal anchors are provided.
Gains concentrate on anomaly identification tasks such as chronological violations, redundancy detection, and contextual robustness, indicating that temporal-order supervision is what unlocks these abilities.
Long-horizon training transfers to conventional benchmarks, with the best results on all bi-temporal change-understanding datasets, so the long-sequence specialization does not sacrifice general remote sensing understanding.
All tested methods degrade as input sequence length grows, but LongEarth-R1 degrades least, retaining 51.4 accuracy on 26-30-frame inputs after reaching 87.8 on 2-5-frame inputs, showing long-context robustness remains the open frontier.
A wider evidence span generally helps LongEarth-R1 when complementary observations support the answer, with one exception at 23-30 frames, so cross-frame evidence integration is a learned skill rather than a simple length penalty.

Reading between the lines

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

Editorial inference: because the model is trained and evaluated on the same benchmark-derived protocol, the headline result is best read as proof of capability on LongEarth-Bench rather than proof of general long-horizon reasoning; a held-out set of unseen Earth observation sequences would test transfer.
Editorial inference: the same decomposition into format, temporal, and spatial rewards could be carried over to other sequence reasoning domains such as video question answering, climate trajectory prediction, or multi-temporal medical imaging, where frame-index grounding is similarly missing.
Editorial inference: if the rule-based labels contain systematic errors in frame ranges or anomaly placements, the temporal and spatial rewards would reinforce those errors; an independent label audit on a random subset would quantify that risk.
Testable extension: varying the number of input frames and the question's evidence span independently would clarify whether LongEarth-R1's robustness comes primarily from sequence grounding, from reward alignment, or from the base model's scale.

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.

Desk Editor's Note

read the letter

First thing you should know: LongEarth-Bench is a genuinely new and useful asset, but the paper's headline numbers are currently not interpretable because the model is trained on the very benchmark it is then evaluated on, with no train/test split described anywhere. That is the elephant in the room.

What is actually new: the benchmark itself. 120k QA samples, average 15.14 frames (max 30), 12 tasks across evolution summarization, spatial reasoning, anomaly identification, and logical prediction, built from five public datasets. That is a real step up from TEOChat's 2-frame average. The 30k structured reasoning subset is a nice addition, even if the traces are machine-generated by Qwen3-VL-8B-Thinking with only spot-check human validation. The training recipe—SFT with explicit sequence IDs plus GRPO with format/temporal/spatial rewards—is sensible and well-motivated, and the ablations are thorough. Table 2 also shows the model transfers reasonably to standard RS benchmarks when fine-tuned on their training splits.

The soft spots, in order of severity. First and most important: there is no train/test split for LongEarth-Bench. The method section says Stage 1 SFT uses LongEarth-Bench samples and Stage 2 GRPO uses the same benchmark; the experiments report accuracy on LongEarth-Bench with no mention of held-out sequences. The baselines marked with * are fine-tuned on LongEarth-Bench too, so everyone is being evaluated on training data. This makes the 'best on all 12 tasks' claim unverifiable. It could be memorization. The authors know how to do proper held-out evaluation—they diligently fine-tune on AID, UCM, etc. training splits—so the omission in the core evaluation looks like an oversight, not a style choice. Second: no error bars or significance tests anywhere. The gaps between LongEarth-R1 and the best baseline on some tasks (e.g., T11: 89.87 vs 87.62) might be noise. Third: the CoT traces are machine-generated; the paper reports automatic verification and human review but no numbers on inter-rater agreement or how often traces were rejected. Fourth: no release plan for data or code is stated.

The benchmark construction itself is careful: diverse sources, rule-based answer generation with human review, and controlled perturbations for anomaly tasks. If the authors release the data with a proper split and show long-horizon gains on held-out sequences, this becomes a solid contribution. As it stands, the central claim is unsupported.

Who is this for: anyone working on RSVLMs or long-sequence multimodal reasoning. It is worth reading as an example of benchmark design, and as a cautionary tale about evaluation.

My recommendation: send it to peer review, but the reviewers should demand a held-out split, error bars, artifact release, and more transparent CoT validation before the numbers are taken seriously.

Referee Report

3 major / 4 minor

Summary. The paper introduces LongEarth-Bench, a large benchmark for long-horizon Earth observation reasoning containing approximately 120k question-answering samples derived from 117k unique images, with an average sequence length of 15.14 frames and a maximum of 30 frames. The benchmark spans 12 tasks across four cognitive dimensions: evolution summarization, spatial reasoning, anomaly identification, and logical prediction, and includes a 30k-sample subset with structured reasoning traces. The paper also presents LongEarth-R1, a vision-language model built on Qwen2.5-VL-7B, trained with supervised fine-tuning using explicit sequence identifiers and structured chain-of-thought supervision, followed by GRPO with format, temporal, and spatial rewards. The authors report that LongEarth-R1 achieves the best results on all 12 long-sequence tasks while remaining competitive on standard remote sensing benchmarks.

Significance. If the results are valid, LongEarth-Bench fills a clear gap in sequence length and cognitive coverage for remote sensing vision-language models, and LongEarth-R1 demonstrates substantial gains on anomaly identification and long-range temporal reasoning. The release of the benchmark and the two-stage training recipe would be a useful contribution to the community. The ablations support the individual contributions of sequence identifiers, CoT supervision, and the proposed rewards. However, the lack of a described train/test split for LongEarth-Bench is a critical flaw that casts the headline results as potentially reflecting memorization; the paper must address this before the claims can be accepted.

major comments (3)

[§5, Table 1 and §4] No train/test split is described for LongEarth-Bench. The method section states that Stage 1 SFT and Stage 2 GRPO train on LongEarth-Bench samples, and the experiments evaluate the same benchmark without any held-out split. In contrast, the standard benchmark evaluations (§5, 'Performance on Standard Remote Sensing Tasks') explicitly fine-tune on the corresponding training split, showing that the authors recognize the need for held-out evaluation. Without a disjoint test split, the claimed 'best results on all 12 long-sequence tasks' is unsubstantiated; it could be an artifact of training-set memorization. Please specify and use a held-out test split for LongEarth-Bench and report results on it, or explicitly justify why the current evaluation is not affected by leakage.
[§3, Dataset Construction] The ground-truth answers are generated by rule-based pipelines and machine-generated CoT traces with only 'rule-based checks', 'human review', and 'spot-check validation'. No quantitative label-quality evidence (e.g., inter-annotator agreement, error rates, number of samples reviewed, or ambiguity resolution) is provided. Since LongEarth-Bench is the sole basis for the claimed state-of-the-art results, label correctness is load-bearing. Please provide a label-quality study with sample sizes, disagreement statistics, and a description of how ambiguous cases were resolved.
[§3, Anomaly tasks (T7–T9)] The controlled perturbations used to construct anomaly tasks (order perturbations, repeated-frame insertions, contextual disturbances) may create non-visual shortcuts. For example, redundancy detection (T8) could be solved by exact pixel matching of repeated frames without any long-horizon reasoning. The paper should demonstrate that models cannot exploit such low-level cues, for instance by adding noise or randomizing frame order at evaluation time, and should report human performance to calibrate the difficulty and validity of these tasks.

minor comments (4)

[Figure 1 and Figure 2] The figures are extremely dense, with small fonts and many overlaid elements; please enlarge and simplify them for readability.
[§4, Spatiotemporal Alignment via GRPO] The reward definitions are said to be in the supplementary material, but no supplementary file is provided; please include the detailed reward formulas and hyperparameters (λ_f, λ_t, λ_s, G, β, ε) in an appendix.
[§5, Experimental Setups] The GRPO hyperparameters (group size G, KL coefficient β, clipping coefficient ε) and the reward weights are not reported in the main text; please provide them for reproducibility.
[Table 3] The row headers in the upper block of the ablation table are ambiguous; consider using a cumulative notation (e.g., '+' for adding each component) to clarify that each row includes all previous components.

Circularity Check

1 steps flagged · score 6.0 of 10

LongEarth-R1 is trained on LongEarth-Bench and then evaluated on the same LongEarth-Bench tasks; the headline 'best on all 12 tasks' is an in-sample fit rather than a held-out prediction.

fitted input called prediction [Method: 'Supervised Sequence Grounding and Reasoning' (Eq. 2) and 'Spatiotemporal Alignment via GRPO' (Eqs. 3-4); Experiments: 'Performance on LongEarth-Bench' (Table 1)]
"Structured CoT Supervision. Within the supervised stage, we use the structured reasoning subset of LongEarth-Bench to supervise evidence-grounded responses. [...] Performance on LongEarth-Bench. We evaluate the 12 LongEarth-Bench tasks spanning EvoSum, Spatial, AnomID, and LogPred. Table 1 shows that LongEarth-R1 ranks first on all tasks."

The model is optimized on LongEarth-Bench reference answers in both training stages: Eq. (2) trains on LongEarth-Bench answer/CoT targets, and the GRPO rewards in Eq. (3) are computed against LongEarth-Bench temporal and spatial ground truth. The evaluation then reports Accuracy, Temporal F1, and Spatial F1 on the same LongEarth-Bench tasks, with no train/test split described anywhere in the paper. Consequently the claimed 'best results on all 12 long-sequence tasks' is an in-sample fit (training-set accuracy), not a prediction of generalization. The value is not forced to be 100%, but its status as a benchmark comparison is undermined because the test answers were used as training targets and the evaluation metrics overlap with the training rewards.

full rationale

The central problem is in-sample evaluation: LongEarth-R1 is trained on LongEarth-Bench and then 'evaluated' on the same 12 tasks, so the headline result reduces to fitting the benchmark's own answers. This is the fitted-input-called-prediction pattern. The paper's external-transfer experiments (AID, UCM, ABCD-CD, CDVQA, xBD, S2Looking, Qfabric, fMoW) are explicitly conducted by fine-tuning on each benchmark's training split and evaluating with the task-native protocol, so those results remain independent evidence and justify score 6 rather than 8 or 10. No load-bearing self-citation chain or uniqueness theorem is present, and the benchmark construction is rule-based from external segmentation/land-cover data, not defined by the model's outputs. The circularity is localized to the LongEarth-Bench evaluation itself, which is exactly where the paper claims state-of-the-art performance on all 12 long-sequence tasks.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claims rest on the benchmark's annotation quality and on the reliability of machine-generated reasoning traces as training targets. The training pipeline introduces tuned reward weights and RL hyperparameters that are not fully specified in the main text.

free parameters (3)

Reward weights (lambda_f, lambda_t, lambda_s) = not reported in main text
Equation (3) combines format, temporal, and spatial rewards with weights; values are deferred to supplementary material and are tuned to balance the objectives.
GRPO group size G = not reported
Number of candidate responses sampled per input in GRPO; affects advantage normalization and is a key hyperparameter.
KL coefficient beta and clipping epsilon = not reported
Equation (4) uses standard PPO-style clipping and KL penalty; values are not given in the main text.

assumptions (4)

domain assumption The four cognitive dimensions and 12 tasks are a valid operationalization of long-horizon Earth observation reasoning.
Introduced by the authors; no external validation that these dimensions cover the space of long-horizon reasoning or that the task labels are meaningful.
domain assumption Rule-based QA generation from segmentation, polygon, land-cover, and flood evidence yields correct, unambiguous answers.
Section 'Dataset Construction' states annotations are generated from these sources; correctness relies on annotation quality and rule-based checks plus human review.
ad hoc to paper Qwen3-VL-8B-Thinking generated CoT traces, after automated checks and human review, are reliable supervision for evidence-grounded reasoning.
Section 'Dataset Construction' states initial traces are generated by this model; fidelity to visual evidence is not quantified in the paper.
ad hoc to paper Explicit text sequence identifiers (Image 1, Image 2, ...) provide stable frame-level temporal anchors for the VLM.
Equation (1) inserts text tokens before each frame's visual tokens; the paper assumes the model binds IDs to frames correctly.

how reviews work

0 comments

Cite this review

Pith. "Pith review of LongEarth-R1: Benchmarking and Aligning Vision-Language Models for Long-Horizon Earth Observation Reasoning." pith.science (2026). https://pith.science/paper/WL57E2XR

@misc{pith2026260813344,
  author       = {Pith},
  title        = {Pith review of: LongEarth-R1: Benchmarking and Aligning Vision-Language Models for Long-Horizon Earth Observation Reasoning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WL57E2XR}},
  note         = {Machine review of arXiv:2608.13344}
}

read the original abstract

Long-horizon Earth observation reasoning requires models to organize multi-stage geographic evolution, localize spatial changes, detect temporal anomalies, and infer future from extended image sequences. However, existing remote sensing vision-language models mainly focus on isolated images, image pairs, or short sequences, limiting reliable grounding in the relevant frames and regions. We introduce LongEarth-Bench, a benchmark containing approximately 120k question-answering samples derived from 117k unique images. Its sequences average 15.14 frames and extend to 30 frames, covering 12 tasks across evolution summarization, spatial reasoning, anomaly identification, and logical prediction. A 30k-sample subset further provides structured reasoning traces linking key frames and changed regions to final answers. We develop LongEarth through supervised fine-tuning with explicit sequence identifiers and structured chain-of-thought supervision. Building on LongEarth, LongEarth-R1 applies group relative policy optimization with format, temporal, and spatial rewards. LongEarth-R1 achieves the best results on all 12 long-sequence tasks while remaining competitive on standard remote sensing benchmarks.

Figures

Figures reproduced from arXiv: 2608.13344 by the authors.

**Figure 1.** Task taxonomy, benchmark positioning, and model capability comparison of LongEarth-Bench and LongEarth-R1. (a) Representative examples of the 12 tasks grouped into four cognitive dimensions. (b) Average sequence length versus cognitive-dimension coverage; bubble size denotes maximum sequence length. (c) Capability coverage of representative remote sensing methods, including long-sequence understanding, evolution de… view at source ↗

**Figure 2.** LongEarth-Bench composition, geographic coverage, and sequence-length statistics. (a) Sample distribution across four cognitive dimensions (inner ring) and 12 tasks (outer ring). (b) Geographic coverage; colors denote the five source datasets and marker size the local mean sequence length. (c) Source-specific length distributions with annotated means. (d) Overall length distribution; bars show counts and the curve g… view at source ↗

**Figure 3.** Construction pipeline of LongEarth-Bench. phenology. As shown in [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Overview of LongEarth and LongEarth-R1. Stage 1 trains LongEarth through sequence-aware supervised fine-tuning with explicit sequence identifiers (Seq. IDs) and structured reasoning traces from the 30k-sample subset. Stage 2 initializes from LongEarth and applies GRPO …

**Figure 5.** Figure 5: Qualitative comparison on long-horizon temporal reasoning. EarthDial and TEOChat predict an incorrect interval, while LongEarth-R1 identifies the correct low-water period and grounds its answer in multi-frame visual evidence. Experimental Results Performance on LongEa…

**Figure 6.** Figure 6: Task-macro scores by input length (top) and ground-truth evidence span (bottom); * denotes LongEarth-Bench fine-tuning. benefit from them. The only exception is the 23–30-frame interval, where TEOChat* attains a higher score. Ablation Analysis of Core Components. We co…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 39 canonical work pages

[1]

AAAI , pages =

Wang, Junjue and Zheng, Zhuo and Chen, Zihang and Ma, Ailong and Zhong, Yanfei , title =. AAAI , pages =

work page
[2]

AAAI , pages =

Pang, Chao and Weng, Xingxing and Wu, Jiang and Li, Jiayu and Liu, Yi and Sun, Jiaxing and Li, Weijia and Wang, Shuai and Feng, Litong and Xia, Gui-Song and He, Conghui , title =. AAAI , pages =

work page
[3]

AAAI , pages =

Li, Yanhui and Cao, Yunkang and Liu, Chengliang and Xiong, Yuan and Dong, Xinghui and Huang, Chao , title =. AAAI , pages =

work page
[4]

AAAI , pages =

Wang, Jiankang and Zhang, Zhihan and Liu, Zhihang and Li, Yang and Ge, Jiannan and Xie, Hongtao and Zhang, Yongdong , title =. AAAI , pages =

work page
[5]

ISPRS Journal of Photogrammetry and Remote Sensing , volume =

Yuan Hu and Jianlong Yuan and Congcong Wen and Xiaonan Lu and Yu Liu and Xiang Li , title =. ISPRS Journal of Photogrammetry and Remote Sensing , volume =

work page
[6]

CVPR , pages =

Kartik Kuckreja and Muhammad Sohail Danish and Muzammal Naseer and Abhijit Das and Salman Khan and Fahad Shahbaz Khan , title =. CVPR , pages =

work page
[7]

IEEE Transactions on Geoscience and Remote Sensing , volume =

Wei Zhang and Miaoxin Cai and Tong Zhang and Yin Zhuang and Xuerui Mao , title =. IEEE Transactions on Geoscience and Remote Sensing , volume =. 2024 , doi =

work page 2024
[8]

IEEE Transactions on Geoscience and Remote Sensing , volume =

Liu, Fan and Chen, Delong and Guan, Zhangqingyun and Zhou, Xiaocong and Zhu, Jiale and Ye, Qiaolin and Fu, Liyong and Zhou, Jun , title =. IEEE Transactions on Geoscience and Remote Sensing , volume =

work page

Show all 47 references

[9]

IEEE Transactions on Geoscience and Remote Sensing , volume =

Zhang, Zilun and Zhao, Tiancheng and Guo, Yulong and Yin, Jianwei , title =. IEEE Transactions on Geoscience and Remote Sensing , volume =
[10]

IEEE Transactions on Geoscience and Remote Sensing , volume =

Chenyang Liu and Rui Zhao and Hao Chen and Zhengxia Zou and Zhenwei Shi , title =. IEEE Transactions on Geoscience and Remote Sensing , volume =. 2022 , doi =

2022
[11]

IEEE Transactions on Geoscience and Remote Sensing , volume =

Chenyang Liu and Keyan Chen and Haotian Zhang and Zipeng Qi and Zhengxia Zou and Zhenwei Shi , title =. IEEE Transactions on Geoscience and Remote Sensing , volume =. 2024 , doi =

2024
[12]

IEEE Transactions on Image Processing , volume =

Yang, Cong and Li, Zuchao and Jiao, Hongzan and Gao, Zhi and Zhang, Lefei , title =. IEEE Transactions on Image Processing , volume =
[13]

Watson and Levente J

Sagar Soni and Akshay Dudhane and Hiyam Debary and Mustansar Fiaz and Muhammad Akhtar Munir and Muhammad Sohail Danish and Paolo Fraccaro and Campbell D. Watson and Levente J. Klein and Fahad Shahbaz Khan and Salman Khan , title =. CVPR , pages =
[14]

ICLR , year =

Jeremy Andrew Irvin and Emily Ruoyu Liu and Joyce Chuyi Chen and Ines Dormoy and Jinyoung Kim and Samar Khanna and Zhuo Zheng and Stefano Ermon , title =. ICLR , year =
[15]

arXiv preprint arXiv:2412.20742 , year =

Yujie Li and Wenjia Xu and Guangzuo Li and Zijian Yu and Zhiwei Wei and Jiuniu Wang and Mugen Peng , title =. arXiv preprint arXiv:2412.20742 , year =

arXiv
[16]

ISPRS Journal of Photogrammetry and Remote Sensing , volume =

Pallavi Jain and Diego Marcos and Dino Ienco and Roberto Interdonato and Tristan Berchoux , title =. ISPRS Journal of Photogrammetry and Remote Sensing , volume =. 2026 , doi =

2026
[17]

Transactions on Machine Learning Research , year =

Zhang, Zhuosheng and Zhang, Aston and Li, Mu and Zhao, Hai and Karypis, George and Smola, Alex , title =. Transactions on Machine Learning Research , year =
[18]

ICLR , year =

Liu, Jiaqi and Sun, Lang and Fu, Ronghao and Yang, Bo , title =. ICLR , year =
[19]

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , volume =

K. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , volume =
[20]

AAAI , pages =

Liang Yao and Fan Liu and Hongbo Lu and Chuanyi Zhang and Rui Min and Shengxiang Xu and Shimin Di and Pai Peng , title =. AAAI , pages =. 2026 , doi =

2026
[21]

Findings of EMNLP , pages =

Yerramilli, Sahiti and Pande, Nilay and Grover, Rynaa and Tamarapalli, Jayant Sravan , title =. Findings of EMNLP , pages =
[22]

arXiv preprint arXiv:2602.07045 , year =

Zhiming Luo and Di Wang and Haonan Guo and Jing Zhang and Bo Du , title =. arXiv preprint arXiv:2602.07045 , year =

arXiv
[23]

IEEE Geoscience and Remote Sensing Magazine , volume =

Weng, Xingxing and Pang, Chao and Xia, Gui-Song , title =. IEEE Geoscience and Remote Sensing Magazine , volume =
[24]

IEEE Geoscience and Remote Sensing Magazine , volume =

Liu, Chenyang and Zhang, Jiafan and Chen, Keyan and Wang, Man and Zou, Zhengxia and Shi, Zhenwei , title =. IEEE Geoscience and Remote Sensing Magazine , volume =
[25]

CVPR , pages =

Van Etten, Adam and Hogan, Daniel and Martinez-Manso, Jesus and Shermeyer, Jacob and Weir, Nicholas and Lewis, Ryan , title =. CVPR , pages =
[26]

Proceedings of the ACM Research in Adaptive and Convergent Systems Conference , year =

Youngsun Jang and Dongyoun Kim and Chulwoo Pack and Kwanghee Won , title =. Proceedings of the ACM Research in Adaptive and Convergent Systems Conference , year =
[27]

CVPR , pages =

Aysim Toker and Lukas Kondmann and Mark Weber and others , title =. CVPR , pages =
[28]

arXiv preprint arXiv:2511.21631 , year =

Shuai Bai and Yuxuan Cai and Ruizhe Chen and others , title =. arXiv preprint arXiv:2511.21631 , year =

arXiv
[29]

arXiv preprint arXiv:2601.04118 , year =

Wenshuai Li and Xiantai Xiang and Zixiao Wen and others , title =. arXiv preprint arXiv:2601.04118 , year =
[30]

IEEE Transactions on Geoscience and Remote Sensing , volume =

Zhenghang Yuan and Lichao Mou and Zhitong Xiong and Xiao Xiang Zhu , title =. IEEE Transactions on Geoscience and Remote Sensing , volume =. 2022 , doi =

2022
[31]

arXiv preprint arXiv:2410.19552 , year =

Hosam Elgendy and Ahmed Sharshar and Ahmed Aboeitta and Yasser Ashraf and Mohsen Guizani , title =. arXiv preprint arXiv:2410.19552 , year =

arXiv
[32]

arXiv preprint arXiv:2411.11360 , year =

Zhiming Wang and Mingze Wang and Sheng Xu and Yanjing Li and Baochang Zhang , title =. arXiv preprint arXiv:2411.11360 , year =

arXiv
[33]

NeurIPS , year =

Weihao Xuan and Junjue Wang and Heli Qi and Zihang Chen and Zhuo Zheng and Yanfei Zhong and Junshi Xia and Naoto Yokoya , title =. NeurIPS , year =
[34]

arXiv preprint arXiv:2502.13923 , year =

Shuai Bai and Keqin Chen and Xuejing Liu and others , title =. arXiv preprint arXiv:2502.13923 , year =

arXiv
[35]

Hu and Yelong Shen and Phillip Wallis and Zeyuan Allen-Zhu and Yuanzhi Li and Shean Wang and Lu Wang and Weizhu Chen , title =

Edward J. Hu and Yelong Shen and Phillip Wallis and Zeyuan Allen-Zhu and Yuanzhi Li and Shean Wang and Lu Wang and Weizhu Chen , title =. ICLR , year =
[36]

Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Xiao Bi and Haowei Zhang and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo , title =. arXiv preprint arXiv:2402.03300 , year =

arXiv
[37]

arXiv preprint arXiv:2305.14467 , year =

Garioud, Anatol and De Wit, Apolline and Poup. arXiv preprint arXiv:2305.14467 , year =

arXiv
[38]

ISPRS Journal of Photogrammetry and Remote Sensing , volume =

Sainte Fare Garnot, Vivien and Landrieu, Loic and Chehata, Nesrine , title =. ISPRS Journal of Photogrammetry and Remote Sensing , volume =
[39]

Wang, Junjue and Xuan, Weihao and Qi, Heli and Liu, Zhihao and Liu, Kunyi and Wu, Yuhan and Chen, Hongruixuan and Song, Jian and Xia, Junshi and Zheng, Zhuo and Yokoya, Naoto , booktitle=
[40]

Shen, Haozhan and Liu, Peng and Li, Jingcheng and Fang, Chunxin and Ma, Yibo and Liao, Jiajia and Shen, Qiaoli and Zhang, Zilun and Zhao, Kangjia and Zhang, Qianqian and Xu, Ruochen and Zhao, Tiancheng , journal =
[41]

IEEE Transactions on Geoscience and Remote Sensing , volume =

Xia, Gui-Song and Hu, Jingwen and Hu, Fan and Shi, Baoguang and Bai, Xiang and Zhong, Yanfei and Zhang, Liangpei and Lu, Xiaoqiang , title =. IEEE Transactions on Geoscience and Remote Sensing , volume =. 2017 , doi =

2017
[42]

Proceedings of the 18th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems , pages =

Yang, Yi and Newsam, Shawn , title =. Proceedings of the 18th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems , pages =. 2010 , doi =

2010
[43]

2017 Fifteenth IAPR International Conference on Machine Vision Applications , pages =

Fujita, Aito and Sakurada, Ken and Imaizumi, Tomoyuki and Ito, Riho and Hikosaka, Shuhei and Nakamura, Ryosuke , title =. 2017 Fifteenth IAPR International Conference on Machine Vision Applications , pages =. 2017 , doi =

2017
[44]

and Doshi, Jigar and Lucas, Keane and Choset, Howie and Gaston, Matthew E

Gupta, Ritwik and Goodman, Bryce and Patel, Nirav and Hosfelt, Ricky and Sajeev, Sandra and Heim, Eric T. and Doshi, Jigar and Lucas, Keane and Choset, Howie and Gaston, Matthew E. , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Wor...
[45]

Remote Sensing , volume =

Shen, Li and Lu, Yao and Chen, Hao and Wei, Hao and Xie, Donghai and Yue, Jiabao and Chen, Rui and Lv, Shouye and Jiang, Bitao , title =. Remote Sensing , volume =. 2021 , doi =

2021
[46]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops , year =

Verma, Sagar and Panigrahi, Akash and Gupta, Siddharth , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops , year =
[47]

Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages =

Christie, Gordon and Fendley, Neil and Wilson, James and Mukherjee, Ryan , title =. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages =. 2018 , doi =

2018

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

AAAI , pages =

Wang, Junjue and Zheng, Zhuo and Chen, Zihang and Ma, Ailong and Zhong, Yanfei , title =. AAAI , pages =

work page

[2] [2]

AAAI , pages =

Pang, Chao and Weng, Xingxing and Wu, Jiang and Li, Jiayu and Liu, Yi and Sun, Jiaxing and Li, Weijia and Wang, Shuai and Feng, Litong and Xia, Gui-Song and He, Conghui , title =. AAAI , pages =

work page

[3] [3]

AAAI , pages =

Li, Yanhui and Cao, Yunkang and Liu, Chengliang and Xiong, Yuan and Dong, Xinghui and Huang, Chao , title =. AAAI , pages =

work page

[4] [4]

AAAI , pages =

Wang, Jiankang and Zhang, Zhihan and Liu, Zhihang and Li, Yang and Ge, Jiannan and Xie, Hongtao and Zhang, Yongdong , title =. AAAI , pages =

work page

[5] [5]

ISPRS Journal of Photogrammetry and Remote Sensing , volume =

Yuan Hu and Jianlong Yuan and Congcong Wen and Xiaonan Lu and Yu Liu and Xiang Li , title =. ISPRS Journal of Photogrammetry and Remote Sensing , volume =

work page

[6] [6]

CVPR , pages =

Kartik Kuckreja and Muhammad Sohail Danish and Muzammal Naseer and Abhijit Das and Salman Khan and Fahad Shahbaz Khan , title =. CVPR , pages =

work page

[7] [7]

IEEE Transactions on Geoscience and Remote Sensing , volume =

Wei Zhang and Miaoxin Cai and Tong Zhang and Yin Zhuang and Xuerui Mao , title =. IEEE Transactions on Geoscience and Remote Sensing , volume =. 2024 , doi =

work page 2024

[8] [8]

IEEE Transactions on Geoscience and Remote Sensing , volume =

Liu, Fan and Chen, Delong and Guan, Zhangqingyun and Zhou, Xiaocong and Zhu, Jiale and Ye, Qiaolin and Fu, Liyong and Zhou, Jun , title =. IEEE Transactions on Geoscience and Remote Sensing , volume =

work page