Writing study: Efficient Unrolled Networks for Large-Scale 3D Inverse Problems
← CVPR 2026 oral writing corpus
Romain Vo, Julián Tachella · Official CVPR 2026 oral record
Figure, table and section ledger
Editorial contribution type: Efficiency, inference & systems (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
- ≈ 8.13 page-equivalents
- Main figures / tables
- 4 / 4
- Attached appendix figures / tables
- 0 / 0
- Attached appendix span
- Separate file recorded below; not in this PDF total
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 | ≈ 1.05 pp. | 9.55% |
| 2. Related work | 2 | ≈ 0.9 pp. | 8.18% |
| 3. Background | 3 | ≈ 0.85 pp. | 7.73% |
| 4. Method | 4 | ≈ 2.26 pp. | 20.55% |
| 5. Experiments | 6 | ≈ 2.05 pp. | 18.64% |
| 6. Limitations | 8 | ≈ 0.36 pp. | 3.27% |
| 7. Conclusion | 8 | ≈ 0.26 pp. | 2.36% |
| Acknowledgements | 9 | ≈ 0.08 pp. | 0.73% |
| References | 9 | ≈ 2.87 pp. | 26.09% |
All detected figure and table locations
Main text
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 4 physical pages; 6 detected figures and 0 tables. These may include overlapping main-text material.
Located appendix only: starts on physical p.1; ≈ 3.94 page-equivalents, 6 figure labels and 0 table labels. This interval is distinct from the whole file.
Argument and organization
- 3. Background and 4. Method
Builds the unrolling setting before partitioning, approximation, and their combination.
- 5. Experiments and 6. Limitations
Tests two inverse-problem families with quality/cost comparisons and ablations.
Problem, claim and contribution
An opening memory/time plot diagnoses the network step as the scaling bottleneck, giving the efficiency contribution a specific target.
Figure 1 p. 1 · Main paper · Additional source 2
Argument and section order
The paper first presents two independently motivated techniques, then combines them; test-time use without ground-truth context is explained separately.
§4 pp. 4–6; Algorithm 1 p. 5 · Main paper · Additional source 2
Figures, captions and visual explanation
A compact vector partition schematic makes the unknown patch and fixed context visible; reconstruction strips add matched zooms across methods.
Figure 2 p. 4; Figure 3 p. 7 · Main paper · Additional source 2
Tables, comparisons and evidence
Quality tables include peak VRAM and training seconds per step, report OOM as infeasibility rather than a fabricated baseline score, and distinguish amplitude-image metrics.
Tables 2–3 pp. 6–7 · Main paper · Additional source 2
Appendix and reproducibility
The supplement sweeps patch size to show quality–memory–time trade-offs and visualizes approximation error beside the learned filter and mask.
Supplement §§A–C; Figures 7–10 · Main paper · Additional source 2
Clarity and writing conventions
The text explicitly notes that refinement helps CBCT more than MRI, limiting a tempting one-size-fits-all benefit claim.
§4.1 p. 4 · Main paper · Additional source 2
What to reuse—and what not to infer
Transferable lesson: An efficiency paper should start with a measured bottleneck and end with paired quality, memory, and time evidence under stated hardware.
Boundary: Training seconds per step, inference latency, and peak memory are distinct quantities; the examined tables do not justify swapping them.
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 main PDF pp. 1–8: framing, section structure, method exposition, figure/table captions, evaluation organization and conclusions; not an exhaustive technical proof audit or reproduction.
- Appendix / supplement
- Scanned the four-page official ZIP-member PDF for train/test configurations, measurement examples, patch-size quality/memory/time sweeps, and normal-operator approximation visualizations; accompanying code was not executed.
- Visually inspected PDF pages
- 4, 7
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.