Bing Wang

—

Unclaimed · AI-structured profilename-matched identity

Computer vision researcher bridging generative image synthesis and LLM-based long-video understanding

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-15) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Output metrics are not yet available: the underlying bibliometric source supplies no publication or preprint counts. This axis would measure research volume and output cadence over time, which cannot be inferred from two cataloged 2026 papers alone.

Impact—

Impact metrics such as citation counts, h-index, and venue-level influence are unavailable for this name-resolved profile. This axis would measure how widely the work is cited and adopted, and it should not be scored from the current snapshot.

Focus0/100

The focus score is 0/100, derived from a field distribution split exactly evenly: 50% computer vision work on long video understanding and multimodal LLMs, 50% on generative models. The two cataloged papers sit one in each subfield, so no dominant research line has consolidated yet. This even split reads as low topical concentration, typical of an early-stage or bridging profile, rather than a quality judgment.

Collaboration41/100

Collaboration scores 41/100, based on seven distinct coauthors recorded in the bibliometric snapshot. The frequent-collaborator list centers on Weixin Xu, Zhenyu Yang, Shengsheng Qian, Changsheng Xu, Yuexing Han, and Haoxuan Zhang, a compact cluster consistent with small-team deep learning research. On both cataloged papers Bing Wang appears neither first nor corresponding — last author on FRPSS and a middle author on VideoScout — suggesting contribution through an established group rather than solo leadership.

Activity—

Activity metrics such as recent publication cadence or yearly output trend are not available. The axis would measure ongoing research tempo over recent years, but the two cataloged papers are both dated 2026, too short a window to score.

Reproducibility38/100

Reproducibility scores 38/100, the mean across the two papers Evolve has assessed (reproPapers: 2). FRPSS reaches medium readiness, reporting best SIFID on three benchmarks but with ablations that never isolate the core Pre-Shape-Space claim; VideoScout is rated low, showing strong open-source gains and thorough ablations yet no significance tests and no released code. Together they place reproducibility below the midpoint.

Research summary

Bing Wang is a computer vision researcher whose reviewed work to date splits evenly between two subfields: generative models for single-image generation and long video understanding built on multimodal large language models. The profile is name-resolved from public bibliographic data and records seven distinct coauthors, with collaboration (41/100) the strongest of the three scored axes. Two 2026 arXiv papers are cataloged. On the generative side, FRPSS (2026) replaces noise initialization in single-image GANs with feature rearrangement in a Pre-Shape Space grounded in Kendall shape theory and geodesic interpolation, reporting the best SIFID on three benchmarks; Evolve notes the ablations never isolate the core claim. On the video side, VideoScout (2026) trains a 7B agent with SFT plus DAPO reinforcement learning to pace its exploration adaptively for long-video question answering, delivering strong open-source gains and thorough ablations, though without significance tests or released code. Author positions — last on FRPSS, middle on VideoScout, neither corresponding — suggest a contributing role within an established group. Reproducibility scores 38/100 across the two papers (medium readiness for FRPSS, low for VideoScout), and the exact 50/50 field split yields a focus score of 0/100. Output, impact, and activity metrics are not yet available and are left unscored.

Collaboration pattern

The snapshot records seven distinct coauthors and a collaboration score of 41/100. Work clusters around a stable group — Weixin Xu, Zhenyu Yang, Shengsheng Qian, Changsheng Xu, Yuexing Han, and Haoxuan Zhang — consistent with a single lab ecosystem. Bing Wang holds neither first nor corresponding authorship on either cataloged paper (last on FRPSS, middle on VideoScout), indicating a contributing-collaborator role within established teams rather than independent project leadership.

Research keywords

single-image generationgenerative adversarial networksPre-Shape SpaceKendall shape theorygeodesic interpolationfeature rearrangementlong video understandingmultimodal large language modelsagentic explorationreinforcement learningDAPOvideo question answering

Highlights

Shape-space feature rearrangement for single-image GANs
FRPSS (2026) replaces noise initialization with feature rearrangement in a Pre-Shape Space built on Kendall shape theory and geodesic interpolation, achieving the best SIFID across three benchmarks; Evolve flags that ablations never isolate the core claim, with medium reproducibility readiness.
Agentic exploration with adaptive pacing for long-video QA
VideoScout (2026) trains a 7B agent with SFT plus DAPO reinforcement learning to adaptively pace its exploration of long videos, delivering strong open-source gains and thorough ablations; Evolve notes the absence of significance tests and released code, with low reproducibility readiness.

Timeline

2026FRPSS appears on arXiv: pre-shape-space feature rearrangement replaces noise initialization in single-image GANs, with Bing Wang listed as last author (non-corresponding).
2026VideoScout appears on arXiv: a 7B agent trained with SFT plus DAPO reinforcement learning learns adaptive exploration pacing for long video understanding, with a middle-author position.
2026Both papers enter the Evolve catalog, one in each CV subfield for an even 50/50 field split, with reproducibility readiness of medium and low respectively.

Papers in the Evolve catalog

Frequent co-authors

Weixin Xu×1 Zhenyu Yang×1 Shengsheng Qian×1 Changsheng Xu×1 Yuexing Han×1 Haoxuan Zhang×1 Bing Wang×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Bibliometric scores derive from the provided public bibliometric snapshot only; no institution, country, or ORCID could be verified for this profile.
Bing Wang is a common name; this profile is name-resolved from public bibliographic data, and the frequent-coauthor list itself contains a same-name entry, so some records may belong to different researchers.
Coverage is limited to two 2026 arXiv preprints; peer-reviewed publications, citation counts, and h-index are outside this data scope.
The reproducibility score (38) rests on only two Evolve-reviewed papers plus author-position inference, and should not be read as a code-availability audit.