Writing study: Seeing Through the Weights: Privacy Leakage in Scene Coordinate Regression
← ECCV 2026 oral writing corpus
Oleksii Nasypanyi, Jaemin Cho, Utku Ozbulak, Byungkon Kang, Francois Rameau · Official ECCV 2026 oral record
Figure, table and section ledger
Editorial contribution type: Empirical analysis, diagnosis & auditing (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
- 18 physical pages
- Before references / appendix
- ≈ 14.67 page-equivalents
- Main figures / tables
- 7 / 3
- Attached appendix figures / tables
- 0 / 0
- Attached appendix span
- No boundary detected; not assumed zero
Extraction notes: No attached appendix boundary detected. Zero means no recognized caption, not independently verified absence.
Source-bound correction: All top-level main headings and the reference boundary checked against source PDF text; source-bound positions use physical single-column y/page-height, not a two-column reading-order surrogate. Running page-number/author titles, affiliations, numbered contribution bullets, footnotes and bibliography continuations are excluded. This linked manuscript contains no appended supplementary section; separate supplement links were not scanned.
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 | ≈ 1.91 pp. | 10.61% |
| 2 Related Works | 3 | ≈ 1.71 pp. | 9.5% |
| 3 Background | 5 | ≈ 1.35 pp. | 7.5% |
| 4 Method | 6 | ≈ 3.72 pp. | 20.67% |
| 5 Experiments | 10 | ≈ 4.88 pp. | 27.11% |
| 6 Conclusion | 15 | ≈ 0.26 pp. | 1.44% |
| 7 Acknowledgments | 15 | ≈ 0.12 pp. | 0.67% |
| References | 15 | ≈ 3.33 pp. | 18.5% |
All detected figure and table locations
Main text
Attached appendix
No recognized caption labels.
Argument and organization
- 1 Introduction; 2 Related Works; 3 Background and Threat Model
Challenges the assumption that storing a scene only in weights removes privacy risk, then defines the adversary's actual access before the attack mechanism.
- 4 Method: querying, stability, refinement, filtering and inversion
Separates black-, gray- and white-box access, describes a full white-box pipeline, and uses experiment headings to vary access, architecture, proxy domain and query budget.
- 5 Experiments; 6 Conclusion
The main result table groups identical metric columns by access level, while the architecture table states different indoor/outdoor distance units and thresholds.
Problem, claim and contribution
Challenges the assumption that storing a scene only in weights removes privacy risk, then defines the adversary's actual access before the attack mechanism.
Abstract and Sections 1 and 3, pp.1–6 · Main paper
Argument and section order
Separates black-, gray- and white-box access, describes a full white-box pipeline, and uses experiment headings to vary access, architecture, proxy domain and query budget.
Sections 3.2–5.5, pp.5–13 · Main paper
Figures, captions and visual explanation
The numbered pipeline maps proxy queries through optional refinement, filtering and accumulation to feature inversion; its branches make added access-dependent operations visible.
Figure 3, p.7 · Main paper
Tables, comparisons and evidence
The main result table groups identical metric columns by access level, while the architecture table states different indoor/outdoor distance units and thresholds.
Tables 1–2, p.11 · Main paper
Appendix and reproducibility
The source points to supplementary pseudo-ground-truth comparisons, domain variants and additional reconstruction examples; those separate results were not inspected.
Sections 5.1–5.5, pp.10–14 · Main paper
Clarity and writing conventions
The experiment setup identifies pseudo-ground-truth construction and discusses its limits instead of implying direct access to every original scene measurement.
Section 5.1, p.10; Section 5.5, pp.13–14 · Main paper
What to reuse—and what not to infer
Transferable lesson: For a privacy or robustness analysis, put the threat model before the mechanism and group comparisons by access, query budget and metric units.
Boundary: Leakage demonstrated in tested models does not quantify every deployment's risk. Reconstruction relative to pseudo-ground truth must not be described as identical to all original training imagery.
Apply this lesson using the argument and evidence worksheet.
Inspection record and sources
- Checked
- 2026-10-02
- Reading status
- Writing-focused manuscript scan + selected visuals inspected
- Main text
- Writing-oriented scan across the manuscript main argument, physical PDF pp.1–15: framing and section roadmaps, method/argument explanations, figure and table captions, comparison protocols, conclusions and limitations. Selected full-page rasters were inspected. This is not exhaustive line-by-line technical reading, proof verification or experimental reproduction.
- Appendix / supplement
- Main-manuscript pointers to appendix/supplement and the reference boundary were checked. Separate supplementary PDFs were not retrieved or inspected; their contents are not included in inspected-page counts.
- Visually inspected PDF pages
- 3, 7, 11
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.