Yingxin: h-index, Total Citations, and Citation Map
Yingxin's h-index is 7 (7 i10-index, 528+ total citations across 17+ publications) according to Google Scholar as of June 2026. Yingxin is affiliated with Graph Origin.
Yingxin is a researcher affiliated with Graph Origin, specializing in MLLM, Deep Learning. Their work has been cited 528 times. This profile visualizes their global influence, highlighting strong citation networks in China.
Yingxin's Citation Metrics
Bibliometric impact based on 17 indexed publications. Of these, 16 are original research articles — the rest are literature highlights, conference abstracts or theses.
- H-Index
- 7
- i10-Index
- 7
- Total Citations
- 528
- Citing Countries
- 34
As of June 2026.
Yingxin has an h-index of 7 and 528 total citations across 17 publications, with research cited by institutions in 34 countries.
Download Exports (PNG, CSV, Poster)
Free Viewing Yingxin's citation map is always free. Pay once to download poster, PNG, and CSV files for offline use or your visa packet.
Global Impact Map
Visualizing the geographic distribution of institutions that have cited your work.
Starting…
Pins will appear here as institutions are resolved — no need to refresh.
A comprehensive survey in llm (-agent) full stack safety: Data, training and deployment
2025136
Top Citing Countries
Top Citing Institutions
Visa Evidence Package
Views and exports tuned for EB-1A, O-1A, and EB-2 NIW petitions. Sustained acclaim, geographic reach, and independent-citation filtering are the strongest evidence categories immigration adjudicators look for.
Significant Contributions
Auto-detected research lines — a seminal paper and the follow-up work building on it. Review and edit before using in a petition. Each Free PDF opens in a new tab — EB-1A organises this into the structure USCIS applies to Criterion 5 of 8 CFR § 204.5(h)(3)(v); EB-1B re-frames it under § 204.5(i)(3) (outstanding researcher); NIW presents it under prong 2 of Matter of Dhanasar.
163 citing papers could not be classified (no author data) — excluded from the percentages above.
The researcher pioneered a segmentation-based approach to deepfake detection and localization, subsequently expanding this framework to address multimodal evaluation and generalized multi-scenario robustness.
The researcher developed a fully open framework for multimodal training, significantly advancing the democratization of access to large-scale vision-language model development.
The researcher established a comprehensive framework for full-stack LLM agent safety, addressing critical gaps across data, training, and deployment phases.
Citation trend (last 10 years)Click to expand
Citation Trend (Last 10 Years)
Related Guides
Learn how to use citation maps for your research and visa applications.
About Yingxin's research
Yingxin is a researcher in MLLM and Deep Learning at Graph Origin. Their work has been cited 528 times across 17 publications (h-index 7), according to Google Scholar.
Their most-cited work, “A comprehensive survey in llm (-agent) full stack safety: Data, training and deployment” (2025), has accumulated 136 citations. Other influential works include “Llava-onevision-1.5: Fully open framework for democratized multimodal training” (2025) with 121 citations and “Shield: An evaluation benchmark for face spoofing and forgery detection with multimodal large language models” (2025) with 69 citations.
Citations of Yingxin's research come primarily from China, United States and Singapore, reflecting international research impact across 5+ countries. The interactive citation map above shows the full geographic distribution of the institutions citing this work.
