Writing study: Removing Cost Volumes from Optical Flow Estimators

ICCV overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← ICCV 2025 oral writing corpus

Simon Kiefhaber, Stefan Roth, Simone Schaub-Meyer · Official ICCV 2025 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: 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

8 / 5

Attached appendix figures / tables

0 / 0

Attached appendix span

No boundary detected; not assumed zero

Before references / appendix: ≈ 73.91%References / intervening material: ≈ 26.09%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	≈ 0.55 pp.	5%
2. Related Work	2	≈ 1.34 pp.	12.18%
3. Enhancing Optical Flow Estimators	3	≈ 2.1 pp.	19.09%
4. Results	5	≈ 3.08 pp.	28%
5. Conclusion	8	≈ 0.29 pp.	2.64%
References	9	≈ 2.87 pp.	26.09%

All detected figure and table locations

Main text

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

Attached appendix

No recognized caption labels.

Argument and organization

1. 1 Introduction; 2 Related Work

   Identifies cost-volume scaling as a concrete memory/compute bottleneck.

2. 3 Enhancing Optical Flow Estimators

   Diagnoses dependence after training, compares fading versus cutting, then selects cutoff schedules and backbones.

3. 4 Results and limitations; 5 Conclusion

   Pairs quality with runtime/memory, evaluates an unseen dataset and then exposes high-resolution failures of the smallest variant.

Problem, claim and contribution

Motivates removal by component-level resource cost and the hypothesis that trained context networks can replace some matching work.

Sections 1 and 3.1; Figure 2, physical pp.1–4 · Main paper

Argument and section order

The study separates whether a cost volume is needed at all from when to remove it and which context backbone to retain.

Tables 1–3 and Section 3.2, pp.4–5 · Main paper

Figures, captions and visual explanation

The architecture distinguishes the removable feature/cost-volume branch from context and refinement; the caption makes the intervention a training strategy rather than just a new box.

Figure 3, p.3 · Main paper

Tables, comparisons and evidence

The main table marks extra training datasets, preserves motion-region breakdowns, and places runtime and memory beside endpoint error for identical full-resolution inputs.

Table 5, p.7; evaluation protocol, pp.5–6 · Main paper

Appendix and reproducibility

Additional resizing-strategy results and complexity breakdowns are delegated to supplemental material; those references, not separate supplemental content, were inspected.

Section 4, pp.6–7 · Main paper

Clarity and writing conventions

Missing points are explicitly attributed to out-of-memory errors, and the limitations name textureless high-resolution holes in the MobileNet variant.

Figure 4, p.6; Section 4.3, p.8 · Main paper

What to reuse—and what not to infer

Transferable lesson: Build an efficiency argument from a bottleneck diagnosis through the training intervention to matched end-to-end quality/resource comparisons.

Boundary: Different backbones and training histories remain visible confounders for causal claims. Batch one, FP32, 10 warmup and 100 timed iterations define the timing protocol; FLOPs alone are not latency.

Apply this lesson using the argument and evidence worksheet.

Inspection record and sources

Checked

2026-10-07

Reading status

Writing-focused manuscript scan + selected visuals inspected

Main text

Writing-oriented scan across the main manuscript, physical PDF pp.1–8, covering framing, section progression, method explanations, captions, comparison protocols and concluding scope; reference boundary checked. Selected full-page rasters were inspected. Not exhaustive line-by-line reading, proof verification or experimental reproduction.

Appendix / supplement

Main-paper appendix/supplement references were checked. A separate supplementary source is linked in the inventory but was not retrieved or inspected; its contents are not included in the reading or visual-page counts.

Visually inspected PDF pages

3, 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.

• Official CVF open-access ICCV 2025 accepted-paper PDF, retrieved 7 October 2026; CVF states identical to accepted version except watermark; not assumed identical to initial submission · 11 pages