Pith. sign in

REVIEW 3 major objections 4 minor 42 references

GRASP: Granularity-Aware Region Alignment and Semantic Prototype Learning for Fine-Grained Cross-Modal Understanding in Drone Views

T0 review · 3 major / 4 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 GRASP claims that object-centric region alignment plus synthesized attribute-level hard negatives, applied only during training, improve fine-grained image-text retrieval in drone views and transfer to unseen aerial scenes.

desk verdict Solid incremental contribution to drone-view retrieval, with an overclaimed zero-shot ERA comparison that needs a controlled baseline before the generalization claim can stand. read the letter →

arxiv 2608.09270 v1 pith:2HBKL4LW submitted 2026-08-10 cs.CV cs.AIcs.IRcs.MM

Jiahui Cui , Yan Zhao , Kan Wei , Enze Zhu , Peirong Zhang , Lei Wang , Yiru Wang This is my paper · ORCID
classification cs.CVcs.AIcs.IRcs.MM
keywords naturallanguage-guideddronescross-modalretrievalfine-grainedunderstandingregion-focusedalignmentsemanticperturbationenhancedmatchingvisualisomorphismzero-shottransferdrone-viewbenchmarks
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

The paper tries to establish a specific diagnosis and cure for fine-grained image-text retrieval from drones. It argues that overhead views create two failure modes: global matching is dominated by background clutter, while top-down geometry makes different buildings look alike, and existing contrastive training does not expose the model to the hard attribute-level cases. It proposes GRASP, which adds Region-Focused Alignment (object-centric contrastive and ranking losses) and Semantic Perturbation Enhanced Matching (synthesized textual and visual hard negatives built from a foreground-purified prototype codebook). On the GeoText-1652 benchmark the framework raises mean recall from 36.05 to 38.04 over the fine-tuned baseline, and when transferred to the unseen ERA dataset it reports mean recall 39.47, above the best fine-tuned ERA method (38.96). Because both additions act only on training losses, the inference pipeline is unchanged.

What carries the argument

The load-bearing machinery is a pair of training-only losses. Region-Focused Alignment computes a region embedding from the attention-weighted patches inside each target bounding box and optimizes a region-level image-text contrastive loss together with a margin-based ranking loss that pushes positives apart from the hardest in-batch negatives. Semantic Perturbation Enhanced Matching synthesizes hard negatives: Textual Semantic Perturbation replaces attribute words (colors, orientations, then general antonyms) through deterministic dictionaries, and Visual Semantic Perturbation builds a Semantic Prototype Codebook, defined as K-Means centroids of attention-selected foreground patches updated by exponential moving average, then interpolates a region feature with a prototype sampled from the top-$k$ nearest neighbors to make a counterfactual visual negative. A binary cross-entropy fine-grained matching loss supervises the fusion encoder over positive, in-batch hard, and synthesized pairs.

What would settle it

Replace the K-Means Semantic Prototype Codebook with the same number of randomly sampled foreground patches (or random in-distribution patches) and retrain; if GeoText-1652 mean recall stays at 38.04 rather than dropping toward the 36.05 baseline, the claim that foreground-purified semantic prototypes drive the visual-perturbation gain is falsified.

Watch

Extended reading notes

Core claim

The central claim is that fine-grained attribute understanding in drone imagery can be improved without any retrieval-time cost by reshaping only the training objectives. Starting from the same XVLM backbone used by the GeoText-1652 baseline, GRASP reports mean recall 38.04 on GeoText-1652 versus 36.05 for the baseline and 36.15 for a prior negative-augmentation method, with the largest gains in image-query R@1 (+3.3). In zero-shot transfer to ERA after training only on GeoText-1652, it reports mean recall 39.47, exceeding the best fine-tuned ERA method (VCSR, 38.96) and improving text-query R@1 by +2.37 over the GeoText-1652 zero-shot baseline. Ablations attribute the gain to both mechanisms, with visual perturbation contributing more than textual perturbation once region-level alignment is present, which the authors read as evidence that visual isomorphism is the harder bottleneck. The paper's stated mechanism is that object-centric region alignment suppresses background, while semantic perturbation forces the model to discriminate subtle attributes such as roof color and materials.

Load-bearing premise

The method assumes that the attention-weighted foreground patches are clean object regions and that interpolating a region feature with a nearby prototype yields valid in-distribution hard negatives; if attention is contaminated by background or prototypes do not correspond to real attributes, the synthesized negatives could mislabel examples and erase the reported gains.

Editorial extensions

If this is right

  • On the paper's numbers, the same backbone reaches higher retrieval accuracy on GeoText-1652 without any change at inference time, so the recipe can be added to existing drone-view retrieval pipelines for free at deployment.
  • The larger improvement on image-query R@1 implies GRASP specifically attacks the visual-isomorphism bottleneck, suppressing geometrically similar but attribute-mismatched buildings.
  • Zero-shot results on ERA suggest the learned attribute discrimination transfers across aerial domains, meaning a model trained on one region can be deployed on a new one without fine-tuning.
  • Because visual perturbation contributes more than textual perturbation, the paper's analysis favors investing in visual attribute-manifold modeling when geometric cues are flattened in overhead views.
  • The method retains a grounding loss for bounding-box supervision, so retrieval quality still depends on the quality of the box annotations used during training.

Reading between the lines

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

  • Implicit extension: if the reported gains are real, the same training-time recipe could transfer to other overhead-imagery retrieval tasks such as satellite-to-text matching or cross-view geolocalization, where background clutter and viewpoint flattening are equally severe; the paper does not test this.
  • Implicit extension: the codebook of foreground prototypes could be inspected as an attribute lexicon (colors, roof materials), offering a built-in explanation channel for why a retrieval succeeded; this interpretability use is not developed in the paper.
  • Implicit extension: because textual perturbation uses deterministic dictionaries, its gains may depend on vocabulary coverage; a direct test would compare dictionary substitution against a language-model-based substitution while controlling for syntactic drift.
  • Implicit extension: the zero-shot result hints that synthesizing hard negatives during training can substitute for target-domain fine-tuning; testing on a third aerial benchmark would show whether that substitution generalizes.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes GRASP, a training-time framework for fine-grained image-text retrieval in drone views, built on an XVLM backbone. GRASP combines Region-Focused Alignment (RFA), which adds region-level contrastive and ranking losses, with Semantic Perturbation Enhanced Matching (SPEM), which constructs textual and visual hard negatives via a Semantic Prototype Codebook (SPC). The authors report fine-tuned mR 38.04 on GeoText-1652 versus 36.05 for the GeoText-1652 baseline, and zero-shot mR 39.47 on ERA, which they claim outperforms the best fine-tuned ERA method (VCSR, mR 38.96). The method adds no inference cost because all new losses are used only during training.

Significance. If the results hold, GRASP would offer a practical contribution to drone-view retrieval by improving attribute-level discrimination without changing inference-time model complexity. The strengths include the use of standard external benchmarks (GeoText-1652 and ERA), an internal ablation study, and a public code release. The two proposed mechanisms are clearly motivated by the drone-view characteristics of background clutter and visual isomorphism. However, the headline zero-shot generalization claim currently rests on an uncontrolled comparison, and the main numbers lack statistical grounding, so the significance of the empirical contribution is not yet fully established.

major comments (3)
  1. [Section 4.5, Table 3] The central claim that zero-shot GRASP (mR 39.47) outperforms the best fine-tuned ERA method (VCSR, mR 38.96) is not supported by the evidence because the comparison is confounded by backbone and pretraining differences. GRASP uses an XVLM backbone pretrained on 16M image-text pairs, whereas the fine-tuned ERA baselines listed (VSE++, PVSE, CLIP, PCME, AMFMN, GALR, VCSR) use different, generally smaller backbones with unknown or different pretraining. This is not an apples-to-apples comparison. The same-backbone zero-shot GeoText-1652 baseline already reaches mR 38.00 on ERA, only 0.96 below VCSR's fine-tuned score, showing that the backbone alone nearly closes the gap. The valid within-backbone comparison is GRASP (39.47) versus GeoText-1652 (38.00), which supports a positive but much smaller claim. The authors should either run a same-backbone fine-tuned ERA baseline or explicitly retract the claim of outperforming all fine-tuned ERA methods.
  2. [Section 4.3, Table 1] The main GeoText-1652 results are reported as single point estimates with no error bars, multiple seeds, or significance tests, yet the paper uses the word 'significant' for the +3.3% R@1 improvement. Given that the mR differences between GRASP and the GeoText-1652 baseline are relatively small (38.04 versus 36.05), measurement noise could affect the conclusions. The authors should report at least three runs with standard deviations or conduct a paired significance test to substantiate the improvement before using 'significant' in the text.
  3. [Section 3.3, Eqs. (12)-(15)] The VSP method relies on a load-bearing assumption that is asserted rather than demonstrated: that attention-derived foreground patch selection (Eq. 12) yields clean object features, that the SPC prototypes span the relevant attribute manifold, and that the convex combination in Eq. (15) stays within the in-distribution manifold. The paper states that this 'guarantees' safety against out-of-distribution noise, but no empirical evidence is provided. If attention weights are contaminated by background or prototypes do not align with semantic attributes, the synthesized negatives could be mislabeled and degrade the fine-grained matching loss (Eq. 18). The authors should provide evidence for the manifold-constrained property, for example by visualizing prototypes, measuring the distance of synthesized features to the training distribution, or ablating with random prototypes instead of SPC-based ones.
minor comments (4)
  1. [Section 3.3, Eq. (8)] The TSP method depends on 'predefined dictionaries' of colors, orientations, and antonyms, but the paper does not describe how these dictionaries were constructed, their size, or their coverage. This information is necessary for reproducibility and should be included in the supplementary material or described in the main text.
  2. [Section 4.2] The two-stage training schedule initializes the SPC using features from the stage-1 model and then fine-tunes with VSP using that same codebook. This self-referential training loop is not discussed; the authors should comment on whether results are sensitive to the quality of the stage-1 model and whether the SPC could be iteratively updated or re-initialized for stability.
  3. [Section 4.4, Table 2] The ablation table is difficult to read because the checkmark columns do not unambiguously map to the configurations described in the text. For example, the sentence 'applying SPEM with semantic perturbations (TSP+VSP) improves performance to 37.01 %' does not clearly match any single row in the rendered table. Please use explicit configuration labels or separate rows for each component combination.
  4. [Section 4.5] The description of the ERA dataset states only that it contains 2,864 drone-view images; the paper should specify the number of text queries and confirm that the official evaluation splits and retrieval protocol are used, so that the zero-shot numbers are interpretable.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: GRASP's gains are measured on held-out GeoText-1652 test splits and an unseen ERA zero-shot benchmark; the SPC self-training loop does not reduce the reported metrics to its own construction.

full rationale

The paper's central results are evaluated on held-out splits of external benchmarks (Section 4.1: GeoText-1652 uses a 600-location test set; ERA is used for zero-shot transfer with no fine-tuning). The reported GRASP figures (Table 1: mR 38.04; Table 3: mR 39.47 on ERA) are not fitted values renamed as predictions: loss weights and hyperparameters are fixed before test evaluation, and no test metric is used to build the Semantic Prototype Codebook or to fit the reported scores. The SPC in Eqs. (12)-(17) is built from the model's own attention outputs and then used to synthesize training-time negatives for the same model; this is a bootstrap-style training loop, but it is not circular in the prohibited sense because the evaluation target is not defined by the codebook construction, and the ERA zero-shot result is an out-of-distribution transfer test. The 'manifold-constrained perturbation' statement in Section 3.3 is a definitional guarantee about convex combinations of in-distribution prototypes, not a derivation of the empirical result; it is an assumption about the training distribution rather than a prediction that reduces to its input. The Table 3 comparison of zero-shot GRASP against fine-tuned ERA baselines is not controlled for backbone or pretraining, so the claim that GRASP outperforms the best fine-tuned ERA method may overstate the contribution of the proposed losses; however, this is a benchmarking-fairness and correctness concern, not circularity, because the baseline numbers are external and are not produced by reusing GRASP's fitted outputs. No load-bearing self-citation chain or imported uniqueness theorem appears; the only overlapping-author reference (MARA, ref [34]) is incidental related work. Accordingly, no circular step can be exhibited with the required quote-and-reduction evidence.

Assumptions & free parameters 11 free parameters · 4 assumptions · 1 invented entities

The central claim rests on a large set of hand-set hyperparameters (loss weights, margins, codebook size, perturbation strengths, and schedule) plus two unproven geometric assumptions about attention-based foreground selection and in-distribution interpolation. The TSP dictionaries are a hidden resource. No physical entities are introduced; the SPC is a model component, not an independently verified object.

free parameters (11)
  • Loss weights lambda1-lambda5 = 1.0, 1.0, 0.6, 0.6, 0.1
    Hand-chosen weights for the five loss terms in Eq. 20; no sensitivity analysis is reported (Section 4.2).
  • Ranking margin m = 0.25
    Margin for the hard-negative ranking loss in Eq. 7 (Section 4.2).
  • SPC codebook size K = 2048
    Number of semantic prototypes in the codebook (Section 4.2).
  • Foreground attention percentile rho = 95%
    Top rho percent of patches by attention are retained as foreground in Eq. 12 (Section 4.2).
  • EMA momentum lambda = 0.995
    Exponential moving average coefficient for codebook updates in Eq. 17 (Section 4.2).
  • VSP mixing coefficient alpha = 0.7
    Interpolation weight for feature-space perturbation in Eq. 15 (Section 4.2).
  • VSP top-k neighbor count = 3
    Number of nearby prototypes considered for sampling in Eq. 14 (Section 4.2).
  • TSP replacement threshold eta = 7
    Target number of attribute words replaced per sentence in the hierarchical TSP (Section 3.3, Section 4.2).
  • Batch size and learning rate = 32, 3e-5
    AdamW optimizer and training schedule used for fine-tuning (Section 4.2).
  • Two-stage epoch schedule = 1 + 2 epochs
    Stage 1 warm-up for SPC initialization, stage 2 with SPEM; no study of this schedule is provided (Section 4.2).
  • TSP attribute and antonym dictionaries = not specified
    The exact color, orientation, and antonym dictionaries are not released; they control the distribution of generated text negatives and are hand-chosen resources (Section 3.3).
assumptions (4)
  • domain assumption Attention weights over bounding-box patches (Eq. 11/12) reliably separate foreground target patches from background clutter.
    VSP constructs the Semantic Prototype Codebook from top-rho percent attention patches; if attention is polluted, prototypes do not represent target attributes.
  • domain assumption Convex combinations of region features and top-k neighbor prototypes stay inside the in-distribution manifold (Eq. 15).
    The paper asserts this guarantee, but no proof or empirical validation is provided.
  • ad hoc to paper The TSP dictionary captures the discriminative attribute vocabulary (colors, orientations, antonyms) in GeoText-1652 captions.
    The exact dictionaries are not released; coverage of test caption attributes is assumed.
  • domain assumption The BCE objective with synthesized negatives does not create false negatives, i.e., perturbed text or features never match a different valid positive.
    Level 3 random substitution can produce captions that could be valid for other images; the paper does not filter for this collision.
invented entities (1)
  • Semantic Prototype Codebook (SPC)
    purpose: Discrete set of foreground-purified attribute prototypes used to synthesize visual hard negatives for fine-grained matching.
    Introduced by the paper; no external falsifiable handle. Its utility is only assessed indirectly through final retrieval metrics.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GRASP: Granularity-Aware Region Alignment and Semantic Prototype Learning for Fine-Grained Cross-Modal Understanding in Drone Views." pith.science (2026). https://pith.science/paper/2HBKL4LW

@misc{pith2026260809270,
  author       = {Pith},
  title        = {Pith review of: GRASP: Granularity-Aware Region Alignment and Semantic Prototype Learning for Fine-Grained Cross-Modal Understanding in Drone Views},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2HBKL4LW}},
  note         = {Machine review of arXiv:2608.09270}
}
read the original abstract

Fine-grained cross-modal understanding in drone views is essential for aerial vision-language navigation. However, the inherent wide field of view and overhead perspective of drone scenarios impose dual challenges on vision-language understanding. At the macro level, overwhelming background clutter in visual representations leads to Cross-Modal Focus Misalignment, where the model prioritizes global environmental similarities over specific object details. At the micro level, Visual Isomorphism creates ambiguity, where candidates share similar geometric structures yet differ only in subtle attributes. To address these challenges, we propose the Granularity-Aware Region Alignment and Semantic Prototype (GRASP) learning framework, enhancing discriminative capability through two synergistic strategies. Specifically, we introduce Region-Focused Alignment (RFA) to promote object-centric cross-modal alignment while suppressing background interference. Concurrently, to tackle visual isomorphism, we propose Semantic Perturbation Enhanced Matching (SPEM), which leverages a foreground-purified Semantic Prototype Codebook (SPC) to construct semantically perturbed negatives for fine-grained semantic discrimination. Extensive experiments on the GeoText-1652 benchmark and the unseen ERA dataset demonstrate that GRASP achieves competitive performance in drone-view fine-grained image-text retrieval, validating its effectiveness for cross-modal understanding in aerial scenarios. Our code implementation is available at https://github.com/UCAS-JC/GRASP.

Figures

Figures reproduced from arXiv: 2608.09270 by the authors.

Figure 1
Figure 1. Dual challenges in drone-view understanding: (a) [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Standard passive in-batch sampling mainly yields [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Overview of GRASP. (a) Encoders produce image and text embeddings, while TSP and SPC-based VSP synthesize hard [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Visual Semantic Perturbation (VSP). (a) Foreground [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 6
Figure 6. Figure 6: Qualitative text-to-image retrieval results ranked [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 5
Figure 5. Figure 5: Visualization of activation maps. (a) Text descrip [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 27 canonical work pages

  1. [1]

    Peter Anderson, Qi Wu, Damien Teney, Jake Bruce, Mark Johnson, Niko Sünder- hauf, Ian Reid, Stephen Gould, and Anton Van Den Hengel. 2018. Vision-and- language navigation: Interpreting visually-grounded navigation instructions in real environments. InProceedings of the IEEE conference on computer vision and pattern recognition. 3674–3683

    2018

  2. [2]

    Pablo Arbelaez, Michael Maire, Charless Fowlkes, and Jitendra Malik. 2010. Con- tour detection and hierarchical image segmentation.IEEE transactions on pattern analysis and machine intelligence33, 5 (2010), 898–916

    work page 2010

  3. [3]

    Gabriele Berton, Riccardo Mereu, Gabriele Trivigno, Carlo Masone, Gabriela Csurka, Torsten Sattler, and Barbara Caputo. 2022. Deep visual geo-localization benchmark. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 5396–5407

    work page 2022

  4. [4]

    Valts Blukis, Yannick Terme, Eyvind Niklasson, Ross A Knepper, and Yoav Artzi

    work page

  5. [5]

    Meghan Chandarana, Erica L Meszaros, Anna Trujillo, and B Danette Allen

    work page

  6. [6]

    Guanlin Chen, Pengfei Zhu, Bing Cao, Xing Wang, and Qinghua Hu. 2023. Cross- drone transformer network for robust single object tracking.IEEE transactions on circuits and systems for video technology33, 9 (2023), 4552–4563

    work page 2023

  7. [7]

    Yen-Chun Chen, Linjie Li, Licheng Yu, Ahmed El Kholy, Faisal Ahmed, Zhe Gan, Yu Cheng, and Jingjing Liu. 2020. Uniter: Universal image-text representation learning. InEuropean conference on computer vision. Springer, 104–120

    2020

  8. [8]

    Meng Chu, Zhedong Zheng, Wei Ji, Tingyu Wang, and Tat-Seng Chua. 2024. Towards natural language-guided drones: GeoText-1652 benchmark with spatial relation matching. InEuropean Conference on Computer Vision. Springer, 213–231

    work page 2024

Show all 42 references
  1. [9]

    Zi-Yi Dou, Yichong Xu, Zhe Gan, Jianfeng Wang, Shuohang Wang, Lijuan Wang, Chenguang Zhu, Pengchuan Zhang, Lu Yuan, Nanyun Peng, et al . 2022. An empirical study of training end-to-end vision-and-language transformers. In Proceedings of the IEEE/CVF conference on computer visi...

    2022

  2. [10]

    Jorge Gago, Cyril Douthe, Rafael E Coopman, Pedro Pablo Gallego, Miquel Ribas- Carbo, Jaume Flexas, Jb Escalona, and Hb Medrano. 2015. UAVs challenge to assess water stress for sustainable agriculture.Agricultural Water Management 153 (2015), 9–19

    2015

  3. [11]

    Octavian Ganea, Gary Bécigneul, and Thomas Hofmann. 2018. Hyperbolic entailment cones for learning hierarchical embeddings. InInternational conference on machine learning. PMLR, 1646–1655

    2018

  4. [12]

    Georgios Georgakis, Karl Schmeckpeper, Karan Wanchoo, Soham Dan, Eleni Miltsakaki, Dan Roth, and Kostas Daniilidis. 2022. Cross-modal map learning for vision and language navigation. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 15460–15470

    2022

  5. [13]

    Yicong Hong, Qi Wu, Yuankai Qi, Cristian Rodriguez-Opazo, and Stephen Gould

  6. [14]

    Baichuan Huang, Deniz Bayazit, Daniel Ullman, Nakul Gopalan, and Stefanie Tellex. 2019. Flight, camera, action! using natural language and mixed reality to control a drone. In2019 International conference on robotics and automation (ICRA). IEEE, 6949–6956

    2019

  7. [15]

    Jinghao Huang, Yaxiong Chen, Shengwu Xiong, and Xiaoqiang Lu. 2024. Visual Contextual Semantic Reasoning for Cross-Modal Drone Image-Text Retrieval. IEEE Transactions on Geoscience and Remote Sensing(2024)

    2024

  8. [16]

    Dong-Wook Kim, Tae-Sun Min, Yoonha Kim, Renato Rodrigues Silva, Hae-Nam Hyun, Ju-Sung Kim, Kyung-Hwan Kim, Hak-Jin Kim, and Yong Suk Chung. 2019. Sustainable agriculture by increasing nitrogen fertilizer efficiency using low- resolution camera mounted on unmanned aerial vehicl...

    2019

  9. [17]

    Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. 2023. Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. InInternational conference on machine learning. PMLR, 19730–19742

    2023

  10. [18]

    Junnan Li, Dongxu Li, Caiming Xiong, and Steven Hoi. 2022. Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation. InInternational conference on machine learning. PMLR, 12888–12900

    2022

  11. [19]

    Junnan Li, Ramprasaath Selvaraju, Akhilesh Gotmare, Shafiq Joty, Caiming Xiong, and Steven Chu Hong Hoi. 2021. Align before fuse: Vision and language repre- sentation learning with momentum distillation.Advances in Neural Information Processing Systems34 (2021), 9694–9705

    2021

  12. [20]

    Arjun Majumdar, Ayush Shrivastava, Stefan Lee, Peter Anderson, Devi Parikh, and Dhruv Batra. 2020. Improving vision-and-language navigation with image- text pairs from the web. InEuropean Conference on Computer Vision. Springer, 259–274

    2020

  13. [21]

    Jun-Ichi Meguro, Kiitirou Ishikawa, Takumi Hasizume, Jun-Ichi Takiguchi, Itsuki Noda, and Mitinori Hatayama. 2006. Disaster information collection into geo- graphic information system using rescue robots. In2006 IEEE/RSJ International Conference on Intelligent Robots and Syste...

    2006

  14. [22]

    Abolfazl Mehbodniya, Julian L Webber, Sathishkumar Karupusamy, et al. 2022. Improving the geo-drone-based route for effective communication and connec- tion stability improvement in the emergency area ad-hoc network.Sustainable Energy Technologies and Assessments53 (2022), 102558

    2022

  15. [23]

    Kim-Anh Nguyen, Maximilian Köper, Sabine Schulte im Walde, and Ngoc Thang Vu. 2017. Hierarchical embeddings for hypernymy detection and directionality. InProceedings of the 2017 Conference on Empirical Methods in Natural Language Processing. 233–243

    2017

  16. [24]

    Avik Pal, Max van Spengler, Guido Maria D’Amely di Melendugno, Alessandro Flaborea, Fabio Galasso, and Pascal Mettes. 2024. Compositional entailment learning for hyperbolic vision-language models.arXiv preprint arXiv:2410.06912 (2024)

    2024 arXiv

  17. [25]

    Yuankai Qi, Zizheng Pan, Yicong Hong, Ming-Hsuan Yang, Anton Van Den Hen- gel, and Qi Wu. 2021. The road to know-where: An object-and-room informed sequential bert for indoor vision-language navigation. InProceedings of the IEEE/CVF International Conference on Computer Vision....

    2021

  18. [26]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. 2021. Learning transferable visual models from natural language supervision. InInternational conference on machine learnin...

    2021

  19. [27]

    Md Tahmid Rashid, Daniel Yue Zhang, and Dong Wang. 2020. Socialdrone: An integrated social media and drone sensing system for reliable disaster response. InIEEE INFOCOM 2020-IEEE Conference on Computer Communications. IEEE, 218–227

    2020

  20. [28]

    Royston Rodrigues and Masahiro Tani. 2022. Global assists local: Effective aerial representations for field of view constrained image geo-localization. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 3871–3879

    2022

  21. [29]

    Paolo Tripicchio, Massimo Satler, Giacomo Dabisias, Emanuele Ruffaldi, and Carlo Alberto Avizzano. 2015. Towards smart farming and sustainable agriculture with drones. In2015 international conference on intelligent environments. IEEE, 140–143

    2015

  22. [30]

    Ivan Vendrov, Ryan Kiros, Sanja Fidler, and Raquel Urtasun. 2015. Order- embeddings of images and language.arXiv preprint arXiv:1511.06361(2015)

    2015 arXiv

  23. [31]

    Tingyu Wang, Zhedong Zheng, Chenggang Yan, Jiyong Zhang, Yaoqi Sun, Bolun Zheng, and Yi Yang. 2021. Each part matters: Local patterns facilitate cross-view geo-localization.IEEE Transactions on Circuits and Systems for Video Technology 32, 2 (2021), 867–879

    2021

  24. [32]

    Yeyuan Wang, Dehong Gao, Lei Yi, Linbo Jin, Jinxia Zhang, Libin Yang, and Xiaoyan Cai. 2025. Enhancing fine-grained vision-language pretraining with negative augmented samples. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 8060–8068

    2025

  25. [33]

    E Wardihani, Magfur Ramdhani, Amin Suharjono, Thomas Agung Setyawan, Sidiq Syamsul Hidayat, Sarono Widodo Helmy, Eddy Triyono, and FIRDANIS Saifullah. 2018. Real-time forest fire monitoring system using unmanned aerial vehicle.Journal of Engineering Science and Technology13, 6...

    2018

  26. [34]

    Hui Wu, Haoquan Zhai, Yuchen Li, Hengyi Cai, Peirong Zhang, Yidan Zhang, Lei Wang, Chunle Wang, Yingyan Hou, Shuaiqiang Wang, et al. 2025. MARA: A Multimodal Adaptive Retrieval-Augmented Framework for Document Question Answering. InProceedings of the 33rd ACM International Con...

    2025

  27. [35]

    Yan Xia, Zhendong Li, Yun-Jin Li, Letian Shi, Hu Cao, João F Henriques, and Daniel Cremers. 2024. UniLoc: Towards universal place recognition using any single modality.arXiv preprint arXiv:2412.12079(2024)

    2024 arXiv

  28. [36]

    Junyan Ye, Honglin Lin, Leyan Ou, Dairong Chen, Zihao Wang, Qi Zhu, Conghui He, and Weijia Li. 2025. Where am i? cross-view geo-localization with natural language descriptions. InProceedings of the IEEE/CVF International Conference on Computer Vision. 5890–5900

    2025

  29. [37]

    Yan Zeng, Xinsong Zhang, and Hang Li. 2021. Multi-grained vision language pre-training: Aligning texts with visual concepts.arXiv preprint arXiv:2111.08276 (2021)

    2021 arXiv

  30. [38]

    Xiao Zhang and Michael Maire. 2020. Self-supervised visual representation learning from hierarchical grouping.Advances in Neural Information Processing Systems33 (2020), 16579–16590. MM ’26, November 10–14, 2026, Rio de Janeiro, Brazil Jiahui Cui et al

    2020

  31. [39]

    Zhedong Zheng, Yunchao Wei, and Yi Yang. 2020. University-1652: A multi-view multi-source benchmark for drone-based geo-localization. InProceedings of the 28th ACM international conference on Multimedia. 1395–1403

    2020

  32. [2017]

    In International Conference on Advances in Computer-Human Interactions

    ’Fly Like This’: Natural Language Interface for UAV Mission Planning. In International Conference on Advances in Computer-Human Interactions. Paper NF1676L-26108

  33. [2019]

    Learning to map natural language instructions to physical quadcopter control using simulated flight.arXiv preprint arXiv:1910.09664(2019)

    2019 arXiv

  34. [2021]

    InProceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition

    Vln bert: A recurrent vision-and-language bert for navigation. InProceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition. 1643– 1653

Pith tools

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