Writing study: SAM 3D Body: Robust Full-Body Human Mesh Recovery

CVPR overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← CVPR 2026 oral writing corpus

Xitong Yang, Devansh Kukreja, Don Pinkus, Taosha Fan, Jinhyung Park, Soyong Shin, Jinkun Cao, Jia-Wei Liu, Nicolás Ugrinovic, Anushka Sagar, Jitendra Malik, Matt Feiszli, Piotr Dollár, Kris Kitani · 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

≈ 8.05 page-equivalents

Main figures / tables

5 / 5

Attached appendix figures / tables

0 / 0

Attached appendix span

Separate file recorded below; not in this PDF total

Before references / appendix: ≈ 73.18%References / intervening material: ≈ 26.82%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	≈ 0.8 pp.	7.27%
2. Related Work	2	≈ 0.34 pp.	3.09%
3. 3DB Model Architecture	2	≈ 1.16 pp.	10.55%
4. Model Training and Inference	4	≈ 0.22 pp.	2%
5. Data Annotation and Mesh Fitting	4	≈ 1.1 pp.	10%
6. Training Datasets	5	≈ 0.41 pp.	3.73%
7. Evaluation	5	≈ 3.04 pp.	27.64%
8. Conclusion	8	≈ 0.22 pp.	2%
References	9	≈ 2.95 pp.	26.82%

All detected figure and table locations

Main text

• Figure 1 · p. 1
• Figure 2 · p. 3
• Figure 3 · p. 4
• Table 1 · p. 5
• Figure 4 · p. 6
• Table 2 · p. 6
• Table 3 · p. 7
• Table 4 · p. 7
• Table 5 · p. 8
• Figure 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 5 physical pages; 5 detected figures and 3 tables. These may include overlapping main-text material.

Located appendix only: starts on physical p.1; ≈ 4.94 page-equivalents, 5 figure labels and 3 table labels. This interval is distinct from the whole file.

Detected heading	PDF page	Estimated span
Supplementary Material	1	≈ 0.1 pp.
1. Evaluating 3DB Prompt Following	1	≈ 0.74 pp.
2. Qualitative results on hand performance	1	≈ 0.78 pp.
3. Evaluating 3D Categorical Performance	2	≈ 0.23 pp.
4. Model Training Details	2	≈ 0.47 pp.
5. Annotation Details	3	≈ 1.43 pp.
6. Limitations	4	≈ 0.73 pp.
References	5	≈ 0.45 pp.

• Table 1 · p. 1
• Table 2 · p. 1
• Figure 1 · p. 2
• Figure 2 · p. 2
• Table 3 · p. 3
• Figure 3 · p. 3
• Figure 4 · p. 4
• Figure 5 · p. 4

Argument and organization

1. Problem and contribution

   SAM 3D Body presents model design and supervision quality as joint contributions. The conclusion returns to multiview capture, synthetic data and actively collected difficult cases, rather than attributing broad robustness solely to an encoder–decoder.

2. Argument sequence

   The eight main sections distinguish architecture, training/inference, annotation/fitting, datasets and evaluation. This deliberate separation gives the data engine its own argument instead of burying it in implementation details.

3. Evidence and verification

   Table 1 lists images/frames, subjects and views for training datasets, with a marker identifying hand-decoder sources. Later tables separate common benchmarks, new test sets, hand estimation and difficult categories; the preference figure supplies wins out of 80 as well as percentages.

Problem, claim and contribution

SAM 3D Body presents model design and supervision quality as joint contributions. The conclusion returns to multiview capture, synthetic data and actively collected difficult cases, rather than attributing broad robustness solely to an encoder–decoder.

Introduction; Sections 5–6; conclusion · Main paper · Additional source 2

Argument and section order

The eight main sections distinguish architecture, training/inference, annotation/fitting, datasets and evaluation. This deliberate separation gives the data engine its own argument instead of burying it in implementation details.

Sections 3–7, main PDF pp.2–8 · Main paper · Additional source 2

Figures, captions and visual explanation

Figure 2 shows the shared image encoder and separate body/hand decoders, with visually distinct prompt and parameter tokens. The nearby notation spells out optional hand crops and prompts, helping readers tell automatic inference from user-guided operation.

Figure 2; Sections 3.1–3.3, PDF p.3 · Main paper · Additional source 2

Tables, comparisons and evidence

Table 1 lists images/frames, subjects and views for training datasets, with a marker identifying hand-decoder sources. Later tables separate common benchmarks, new test sets, hand estimation and difficult categories; the preference figure supplies wins out of 80 as well as percentages.

Table 1, PDF p.5; Tables 2–5 and Figure 5, PDF pp.6–8 · Main paper · Additional source 2

Appendix and reproducibility

The supplement tests keypoint and mask prompting, then documents hand/category results, training and manual/automatic/multiview annotation before limitations. This verifies the promptability claim and exposes the work behind the data engine.

Supplementary Sections 1–6, PDF pp.1–5 · Main paper · Additional source 2

Clarity and writing conventions

The evaluation states that MHR meshes are mapped to SMPL for SMPL-based benchmarks and distinguishes backbone variants. Naming this conversion and model identity prevents a broad model-family claim from obscuring the actual comparison.

Section 7, main PDF p.5 · Main paper · Additional source 2

What to reuse—and what not to infer

Transferable lesson: Give model, supervision pipeline and dataset diversity distinct sections; test user-controllable features and difficult subgroups instead of relying on one aggregate benchmark.

Boundary: Human-preference percentages depend on the stated sample and alternatives. Representation conversion and optional prompts must remain visible when comparing accuracy.

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

Writing-oriented scan throughout main PDF pp.1–8 and references pp.9–11: abstract/framing, section sequence, figure/table captions, evaluation organization and conclusions; not an exhaustive technical verification.

Appendix / supplement

Writing-oriented supplement scan: all five supplementary PDF pages, with prompt-following studies, hand/category evaluations, training, annotation and limitations. Video/code assets were not executed or assessed.

Visually inspected PDF pages

3, 5

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 CVPR 2026 accepted-paper PDF · 11 pages
• Official CVF supplementary PDF · 5 pages