Writing study: GeoViS: Geospatially Rewarded Visual Search for Remote Sensing Visual Grounding

CVPR overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← CVPR 2026 oral writing corpus

Peirong Zhang, Yidan Zhang, Luxiao Xu, Jinliang Lin, Zonghao Guo, Fengxiang Wang, Xue Yang, Kaiwen Wei, Lei Wang · Official CVPR 2026 oral record

Observed writing choices, not acceptance rules. This study examines the inspected manuscript version. It does not establish why the paper was accepted or that its presentation choices should be copied unchanged.

Figure, table and section ledger

Editorial contribution type: Architecture & representation (primary).

Layout extraction, not a required paper template. Counts are distinct caption labels; a multi-panel figure counts once. Section spans are estimated from heading positions, including floats and whitespace. Method and limitations.

Canonical PDF

11 physical pages

Before references / appendix

≈ 8.11 page-equivalents

Main figures / tables

4 / 5

Attached appendix figures / tables

0 / 0

Attached appendix span

Separate file recorded below; not in this PDF total

Before references / appendix: ≈ 73.73%References / intervening material: ≈ 26.27%Estimated layout envelopes, not text/artwork area. Unresolved appendix boundaries remain within the undivided post-body material.

Extraction notes: No attached appendix boundary detected. Zero means no recognized caption, not independently verified absence.

Section-by-section lengths and source pages

The title/abstract/front matter precede the first numbered heading. A span ends at the next detected top-level heading, not at the last sentence; end-page whitespace is included. References and statements are listed separately. These are not measurements of text-only or artwork area.

Top-level heading	Starts on PDF page	Estimated span	Share of entire PDF
1. Introduction	1	≈ 1.33 pp.	12.09%
2. Related Work	2	≈ 0.49 pp.	4.45%
3. Methodology	3	≈ 2.53 pp.	23%
4. Experiments	5	≈ 3.1 pp.	28.18%
5. Conclusion	8	≈ 0.2 pp.	1.82%
6. Acknowledgments	9	≈ 0.06 pp.	0.55%
References	9	≈ 2.89 pp.	26.27%

All detected figure and table locations

Main text

• Figure 1 · p. 1
• Figure 2 · p. 4
• Table 1 · p. 6
• Figure 3 · p. 7
• Table 2 · p. 7
• Table 3 · p. 8
• Figure 4 · p. 8
• Table 4 · p. 8
• Table 5 · p. 8

Attached appendix

No recognized caption labels.

Separate source files—excluded from canonical-PDF totals

A separate author version may repeat main text and references. Whole-file lengths and counts are not extra appendix material and are never added to the canonical PDF totals. Read the study's source scope for which sections received substantive reading versus layout inspection.

Additional source 1

Separate PDF (or ZIP containing it): whole-file layout of 8 physical pages; 0 detected figures and 4 tables. These may include overlapping main-text material.

Located appendix only: starts on physical p.1; ≈ 7.93 page-equivalents, 0 figure labels and 4 table labels. This interval is distinct from the whole file.

Detected heading	PDF page	Estimated span
Supplementary Material	1	≈ 0.02 pp.
6. Detailed Visual Search Process	1	≈ 3.7 pp.
7. Atomic Operation Dataset Construct	4	≈ 2.54 pp.
8. Generalization Ability	7	≈ 0.49 pp.
9. Limitation and Future Work	7	≈ 0.44 pp.
References	8	≈ 0.74 pp.

• Table 6 · p. 4
• Table 7 · p. 4
• Table 8 · p. 7
• Table 9 · p. 8

Argument and organization

1. From one-shot grounding to search

   Define the scale/context problem, then formalize the search state, actions and reward.

2. Shared model and separated tests

   Describe VisualRAG's action/reward/grounding roles before benchmark, resolution, operation and transfer studies.

Problem, claim and contribution

The paper motivates sequential search with two concrete remote-sensing obstacles: small targets in large images and queries involving multiple spatial relations.

Introduction and Figure 1, PDF pp. 1–2 · Main paper · Additional source 2

Argument and section order

State/action/reward definitions precede the conditional grounding stage, and the shared VisualRAG model is described separately from the search algorithm.

Sections 3.1–3.4, PDF pp. 3–5 · Main paper · Additional source 2

Figures, captions and visual explanation

The main results table groups specialist grounding models, general VLMs and remote-sensing MLLMs rather than mixing them into an undifferentiated rank list; missing entries remain dashes.

Table 1, PDF p. 6; visual inspection · Main paper · Additional source 2

Tables, comparisons and evidence

Effective-resolution, atomic-operation and cross-dataset tests address different explanations for improvement, while the caption distinguishes externally sourced evaluation results.

Tables 1 and 3–5, PDF pp. 6–8 · Main paper · Additional source 2

Appendix and reproducibility

The supplement unpacks search operators, reward motivation, structured extraction prompts and atomic-operation datasets, including depth/simulation trade-offs.

Supplement Sections 6–7, PDF pp. 1–8 · Main paper · Additional source 2

Clarity and writing conventions

The overview enumerates the shared model's three duties before presenting the MCTS equations, making the source of actions and rewards explicit.

Section 3.1, PDF p. 3 · Main paper · Additional source 2

What to reuse—and what not to infer

Transferable lesson: For a search-augmented VLM, define who proposes actions, who judges them and who emits the final answer; test resolution gains separately from reasoning gains.

Boundary: The specialist rows remain stronger on some metrics, so the claim is not universal best performance over every grounding model. Search depth and simulation count are part of the inference budget.

Apply this lesson using the argument and evidence worksheet.

Inspection record and sources

Checked

2026-09-26

Reading status

Writing-focused manuscript scan + selected visuals inspected

Main text

Manuscript-wide writing scan of PDF pp. 1–11: framing, argument sequence, section organization, figure/table captions, evaluation interpretation and limitations; reference sections identified. This is a writing-focused scan, not an exhaustive technical proof or reproduction audit. Main PDF pages 3 and 6 were rendered and visually inspected.

Appendix / supplement

Writing-focused scan of the separate supplementary PDF, pp. 1–8, including its organization, captions, protocol/evidence detail and limitation material where present; proofs and code listings were inspected for exposition, not independently verified or executed. Supplement Sections 6–7, PDF pp. 1–8. Linked video/audio demonstrations and non-PDF archive contents were not assessed.

Visually inspected PDF pages

3, 6

This is a writing-focused scan of the manuscript and the stated supplementary scope: framing, organization, visual explanation, evidence presentation and wording. It is not a line-by-line proof audit, reproduction of results or exhaustive scientific review.

• Official CVPR 2026 CVF open-access accepted manuscript; retrieved 2026-09-26 · 11 pages
• Official CVPR 2026 CVF supplementary PDF; retrieved 2026-09-26 · 8 pages