Xudong Zhao

—

Unclaimed · AI-structured profileOpenAlex

Computer vision researcher in cross-view geo-localization and label-efficient visible-infrared object detection.

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

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—

Insufficient data to score. This axis would measure total research output — number of publications, venue types, and yearly productivity — but the current snapshot records only two 2026 arXiv preprints, with no full publication history to aggregate.

Impact—

Insufficient data to score. This axis would measure scholarly influence — citation counts, h-index, and citations per paper — but no citation or bibliometric impact figures exist in the current profile, so no estimate is offered.

Focus0/100

Scores 0 because the reviewed output splits evenly: 50% on cross-view geo-localization / remote sensing and 50% on general computer vision (visible-infrared detection). The two lines are technically complementary — one centered on localization, the other on detection — but, on the recorded field distribution, no single dominant direction has emerged yet.

Collaboration55/100

Scores 55. The cataloged papers involve 15 distinct coauthors, with 12 recurring collaborators including Qi Ming, Xuyu Fan, Zhu Han, Liuqian Wang, Siyuan Cao, and Jiahuan Zhou. On both 2026 preprints the researcher sits in a middle author position rather than leading, consistent with a collaborative contributor within a sizeable computer vision team.

Activity—

Insufficient data to score. This axis would measure research tempo — publication frequency, preprint-to-publication cadence, and sustained output across recent years — but only two 2026 records are available, so year-over-year activity cannot be assessed.

Reproducibility55/100

Scores 55, the mean readiness across the 2 papers assessed, each rated medium. MVLGeo's mixture-of-experts plus vision-language reranking pipeline implies nontrivial compute and integration effort, with VLM costs unaccounted; the semi-supervised RGB-IR framework reports unmatched baselines and single-run peaks, so its headline gains warrant independent replication.

Research summary

Xudong Zhao is a computer vision researcher whose cataloged output to date — two 2026 arXiv preprints, both with middle-author positions — spans two complementary threads: cross-view object geo-localization and label-efficient multimodal detection. The first, MVLGeo, unifies UAV and street-view cross-view object geo-localization in a single multi-view mixture-of-experts model, adding vision-language reranking and SAM-derived ellipse priors; Evolve's review notes strong UAV gains, while street-view state-of-the-art coverage is partial and VLM pipeline costs are unaccounted. The second, Aligned Consensus Teaching, is presented as the first semi-supervised framework for image-pair-level visible-infrared oriented detection, combining mean-teacher self-training, pseudo labels, and cross-modal alignment; results are strong but rest on unmatched baselines and single-run peaks. The collaboration record shows 15 distinct coauthors, including 12 recurring partners such as Qi Ming, Xuyu Fan, and Zhu Han, with a collaboration score of 55. Reproducibility averages 55 across the two papers assessed, both rated medium readiness. Focus scores 0, reflecting an even 50/50 split between geo-localization/remote sensing and general computer vision. Output, impact, and activity metrics are not yet available, and with no ORCID or affiliation on record, this profile is name-resolved from public bibliographic data.

Collaboration pattern

The 2026 record shows a team-based working pattern: 15 distinct coauthors across 2 preprints, with 12 recurring collaborators — including Qi Ming, Xuyu Fan, Zhu Han, Liuqian Wang, Siyuan Cao, Xiaohan Zhang, Jiahuan Zhou, and Jiangmeng Li — suggesting a stable core group. On both papers the researcher holds a middle author position without a corresponding-author role, indicating contribution as a co-developer rather than project lead so far.

Research keywords

cross-view object geo-localizationmixture-of-expertsvision-language rerankingUAV / drone navigationremote sensingSAM (Segment Anything) priorsanisotropic Gaussian priorsemi-supervised object detectionvisible-infrared fusionoriented object detectionmean-teacher self-training and pseudo labelscross-modal alignment

Highlights

Unified UAV and street-view object geo-localization with mixture-of-experts
In MVLGeo (2026), the researcher contributes to a multi-view mixture-of-experts model with vision-language reranking that handles UAV and street-view cross-view object geo-localization in one framework, adding SAM-derived ellipse priors; UAV gains are strong, while street-view state-of-the-art coverage is partial and VLM pipeline costs remain unaccounted.
First semi-supervised framework for image-pair-level RGB-IR detection
Aligned Consensus Teaching (2026) introduces the first semi-supervised approach to oriented object detection in weakly-aligned visible-infrared imagery, using mean-teacher self-training and pseudo labels with cross-modal alignment; results are strong, though unmatched baselines and single-run peaks temper the headline claims.
An emerging 2026 profile bridging localization and detection
The two 2026 preprints split evenly (50/50) between cross-view geo-localization / remote sensing and general computer vision, share 15 distinct coauthors, and both carry medium reproducibility readiness — an early-stage, team-based record whose citation-level impact is not yet measurable.

Timeline

2026A first cataloged preprint appears on arXiv: MVLGeo, a multi-view mixture-of-experts model with vision-language reranking for cross-view object geo-localization.
2026A second cataloged preprint appears on arXiv: Aligned Consensus Teaching, a semi-supervised framework for label-efficient oriented detection in weakly-aligned visible-infrared imagery.
2026By year-end, the profile records 2 arXiv preprints, 15 distinct coauthors, and an even 50/50 split between geo-localization/remote sensing and general computer vision.

Papers in the Evolve catalog

Frequent co-authors

Qi Ming×2 Xuyu Fan×1 Zhu Han×1 Liuqian Wang×1 Siyuan Cao×1 Xiaohan Zhang×1 Mingjing Zhao×1 Yuhan Zhang×1 Xiaxin Yuan×1 Jiahuan Zhou×1 Jiangmeng Li×1 Zhanchao Huang×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.

Profile is name-resolved from public bibliographic data; "Xudong Zhao" is a common name and no ORCID, institution, or country is available to confirm identity.
Bibliometric scores derive from a limited OpenAlex-based snapshot; output, impact, and activity axes lack sufficient data and are left unscored.
Both catalog papers are 2026 arXiv preprints; peer-review status and final publication venues are unknown.
The reproducibility score (55) rests on only 2 assessed papers, both rated medium — a small evidence base.