Gait-focused computer vision researcher working on occlusion robustness and language-guided biometric recognition.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
17
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-19) 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 volume data is insufficient for scoring; this axis would measure total indexed works, venue quality, and productivity trends. The current snapshot contains only two Evolve-reviewed papers and no per-venue output list, so no reliable output score can be computed yet.
Impact—
Impact data is insufficient for scoring; this axis would measure citation counts, citation velocity, and influence beyond the immediate subfield. No per-work citation totals are provided in the snapshot, so no estimate or score is offered here.
Focus0/100
The focus score registers 0 because indexed output splits evenly between 'computer vision' (0.5) and 'gait recognition / biometrics' (0.5), two adjacent labels. Raw topic counts nonetheless show concentration: 38 of 100 topic-assigned works concern gait recognition and 26 concern pose and action recognition, indicating a coherent gait-and-motion-centered agenda rather than scattered output.
Collaboration46/100
Collaboration scores 46, a moderate level. The recent window lists nine distinct coauthors, a relatively compact network. Recurring names include Saihui Hou, Xu Liu, and Yongzhen Huang — each appearing twice in the frequent-coauthor list — alongside Panjian Huang, Yunjie Peng, and Zhiqiang He. Both catalog papers position him as a middle author without corresponding role, consistent with a senior contributor inside a small, stable gait-recognition team.
Activity93/100
Activity scores 93, the strongest axis. OpenAlex counts by year show sparse early output — single works in 2004, 2008, 2010, and 2013, then 1–3 works per year through 2022 — followed by a steep rise: 16 works in 2023 and 12 in 2024, with 23 works in the most recent three-year window, roughly a third of all indexed output. This indicates a highly productive current phase.
Reproducibility55/100
Reproducibility scores 55, derived from Evolve reviews of two papers rated 'medium' readiness (mean 55). The occluded-gait mixture-of-experts paper reports strong results but thin margins without significance testing; the vocabulary-guided paper shows consistent but modest gains (+0.5–4%) at roughly doubled parameters. Releasing code, training details, and significance analyses would strengthen verifiability.
Research summary
Chunshui Cao is a computer vision researcher affiliated with World Water Watch (CN), whose OpenAlex-resolved record centers on gait recognition and biometrics. Topic assignments show 38 works on gait recognition and analysis, 26 on human pose and action recognition, 16 on diabetic foot ulcer assessment, and 10 each on video surveillance/tracking and hand gesture recognition. Reported indicators include an h-index of 17 and an i10-index of 27, with a markedly rising activity profile: yearly counts climbed from single digits through 2022 to 16 works in 2023 and 12 in 2024, yielding 23 works in the most recent three years. The field distribution splits evenly between general computer vision and gait-recognition biometrics. Two 2026 arXiv papers reviewed by Evolve — a mixture-of-experts model for occluded gait recognition introducing the OccGait dataset, and a vocabulary-guided approach using LLM-generated gait-phase text embeddings — both list him as a middle author, and both carry 'medium' reproducibility readiness with a mean score of 55, reviewers noting absent significance testing. Collaboration metrics are moderate: nine distinct coauthors, with Saihui Hou, Xu Liu, and Yongzhen Huang recurring most frequently. Overall, the record depicts a steadily intensifying, gait-centered agenda bridging silhouette-based recognition and vision-language methods.
Collaboration pattern
With nine distinct recent coauthors and a collaboration score of 46, Chunshui Cao works within a compact, stable team rather than a broad network. Saihui Hou, Xu Liu, and Yongzhen Huang recur most often (each listed twice), joined by Panjian Huang, Yunjie Peng, and Zhiqiang He. In both cataloged papers he appears as a middle author without corresponding role, suggesting contribution within a gait-recognition group anchored by recurring partners.
Research keywords
Highlights
GaitMoE recasts occluded gait recognition as action detection
Language-guided gait phases yield consistent but modest gains
Sharp publication surge since 2023
Yearly indexed output jumped from 7 works in 2022 to 16 in 2023 and 12 in 2024, with 23 works in the most recent three-year window — driving an activity score of 93.
Timeline
- 2004Earliest indexed work appears in the OpenAlex record; output remains sparse, with single works in 2004, 2008, 2010, and 2013.
- 2020Output begins a sustained climb: 4 works in 2020, 5 in 2021, and 7 in 2022, foreshadowing the later surge.
- 2023Peak productivity: 16 works indexed in a single year, the highest yearly count in the record.
- 2026Two arXiv gait-recognition papers enter the Evolve catalog — GaitMoE for occluded gait and a vocabulary-guided approach — both with middle-author placement.
Papers in the Evolve catalog
Frequent co-authors
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 indicators (h-index 17, i10-index 27, counts by year, topics) come from an OpenAlex-resolved identity and may omit works or conflate same-name authors.
- Only two papers carry full Evolve reviews; quality judgments beyond these rest on raw topic counts alone.
- Output and impact dimension scores are missing from the snapshot; no per-work citation totals were provided.
- The affiliation (World Water Watch, CN) is recorded as given in the source data and was not independently verified.