Writing study: Register Any Point: Scaling 3D Point Cloud Registration by Flow Matching
← ECCV 2026 oral writing corpus
Yue Pan, Tao Sun, Liyuan Zhu, Lucas Nunes, Iro Armeni, Jens Behley, Cyrill Stachniss · Official ECCV 2026 oral record
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
- 22 physical pages
- Before references / appendix
- ≈ 15.15 page-equivalents
- Main figures / tables
- 6 / 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 | 2 | ≈ 1.43 pp. | 6.5% |
| 2 Related Work | 3 | ≈ 1.05 pp. | 4.77% |
| 3 Generative Point Cloud Registration | 4 | ≈ 3.69 pp. | 16.77% |
| 4 Experimental Evaluation | 8 | ≈ 6.84 pp. | 31.09% |
| 5 Conclusion | 15 | ≈ 0.43 pp. | 1.95% |
| Acknowledgements | 15 | ≈ 0.46 pp. | 2.09% |
| References | 16 | ≈ 6.85 pp. | 31.14% |
All detected figure and table locations
Main text
Attached appendix
No recognized caption labels.
Argument and organization
- 1 Introduction; 2 Related Work
Motivates the limitations of composing pairwise registrations and presents multi-view alignment as one conditional generation problem.
- 3 Generative Point Cloud Registration
States the input/output problem before flow matching, rigidity-enforcing inference and canonicalization; evaluation distinguishes established pairwise tests from a cross-domain multi-view benchmark.
- 4 Experimental Evaluation; 5 Conclusion
Registration success is defined separately for edges and whole graphs, with stated translation/rotation thresholds; runtime is broken down by view and token count.
Problem, claim and contribution
Motivates the limitations of composing pairwise registrations and presents multi-view alignment as one conditional generation problem.
Abstract and Sections 1–2, pp.1–4 · Main paper
Argument and section order
States the input/output problem before flow matching, rigidity-enforcing inference and canonicalization; evaluation distinguishes established pairwise tests from a cross-domain multi-view benchmark.
Sections 3–4, pp.4–14 · Main paper
Figures, captions and visual explanation
The overview tracks unordered colored point sets through keypoints and flow generation to recovered rigid poses, explicitly retaining the final geometric recovery stage.
Figure 2, p.5 · Main paper
Tables, comparisons and evidence
Registration success is defined separately for edges and whole graphs, with stated translation/rotation thresholds; runtime is broken down by view and token count.
Section 4.3, p.11; Table 2, p.12; Figure 6, p.14 · Main paper
Appendix and reproducibility
The main text assigns stricter thresholds, benchmark curation and low-overlap results to supplementary material. These pointers were read; the separate supplement was not.
Sections 4.2–4.3, pp.9,11 · Main paper
Clarity and writing conventions
Explicitly states that the flow alone does not guarantee per-view rigidity, then supplies a dedicated inference subsection rather than concealing the postprocessing assumption.
Section 3.3, p.6 · Main paper
What to reuse—and what not to infer
Transferable lesson: For geometric generative methods, show what is predicted and how it becomes a valid geometric output; match robustness plots and success definitions to the deployment claim.
Boundary: Pairwise and multi-view success rates are not interchangeable, and the runtime analysis concedes weaker two-view efficiency before showing scaling benefits.
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
- 4, 5, 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.