Writing study: Chain-of-Visual-Thought: Teaching VLMs to See and Think Better with Continuous Visual Tokens
← ECCV 2026 oral writing corpus
Yiming Qin, Bomin Wei, Jiaxin Ge, Konstantinos Kallidromitis, Stephanie Fu, Trevor Darrell, XuDong Wang · 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
- 20 physical pages
- Before references / appendix
- ≈ 14.37 page-equivalents
- Main figures / tables
- 9 / 6
- 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.
All detected figure and table locations
Main text
Attached appendix
No recognized caption labels.
Argument and organization
- 1 Introduction; 2 Related Work
Contrasts text-only reasoning and external tool calls with an internal continuous visual token chain, making dense perception the targeted deficiency.
- 3 Chain-of-Visual-Thought: tokens, alignment and training
Introduces the token mechanism, token-family choice and four-stage formatting before evaluation; qualitative decoding and matched token controls accompany benchmark results.
- 4 Experiments, ablations and evidence alignment; 5 Conclusions
Zero-token and empty-token controls distinguish ordinary fine-tuning and additional token capacity from aligned visual information; alignment methods are compared separately.
Problem, claim and contribution
Contrasts text-only reasoning and external tool calls with an internal continuous visual token chain, making dense perception the targeted deficiency.
Figure 1 and Section 1, pp.1–3 · Main paper
Argument and section order
Introduces the token mechanism, token-family choice and four-stage formatting before evaluation; qualitative decoding and matched token controls accompany benchmark results.
Sections 3–4, pp.6–14 · Main paper
Figures, captions and visual explanation
The training schematic draws separate expert-supervision paths and the caption explicitly says dense decoding is optional at inference, preventing training-only teachers from appearing required online.
Figure 3, p.5 · Main paper
Tables, comparisons and evidence
Zero-token and empty-token controls distinguish ordinary fine-tuning and additional token capacity from aligned visual information; alignment methods are compared separately.
Tables 5–6 and Section 4.5, p.13 · Main paper
Appendix and reproducibility
The official linked manuscript was inspected through its main argument and references; no separately linked supplementary PDF was retrieved for an appendix study.
Main PDF pp.1–20; official oral event 6085 · Main paper
Clarity and writing conventions
Captions explain the role of each expert and token family, while the method roadmap announces the order of token selection, alignment and loss specification.
Figure 3, p.5; Section 3, p.6 · Main paper
What to reuse—and what not to infer
Transferable lesson: A multimodal architecture diagram should distinguish supervision from inference and pair latent-token visualizations with capacity-matched and supervision-matched controls.
Boundary: Training uses dense expert signals and LoRA adaptation. Tool-free inference should not be rewritten as no additional training supervision or no optional decoding cost.
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
- 5, 13
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.