PeopleofSuperintelligence

Researcher

Sheng-Yu Huang

Sheng-Yu Huang is a Research Scientist at NVIDIA Research Taiwan who works on 3D vision, from point cloud completion to 3D scene understanding. After a B.S. in Electrical Engineering from National Taiwan University in 2019, he joined the Vision and Learning Lab, advised by Prof. Yu-Chiang Frank Wang. His papers include SPoVT at NeurIPS 2022, GSNeRF at CVPR 2024, TPA3D at ECCV 2024, and a 3D Gaussian inpainting method at CVPR 2025. He completed his Ph.D. in 2026 and published OpenVoxel at CVPR 2026, a training-free method for grouping and captioning sparse voxels.

Now at
NVIDIA
Works
2
Citations
4
h-index
1
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2019

    Engineering degree at National Taiwan University

    A B.S. in Electrical Engineering from National Taiwan University comes in 2019. He then joins the Vision and Learning Lab, advised by Prof. Yu-Chiang Frank Wang.

  2. 2022

    SPoVT point cloud completion

    He coauthors SPoVT with Hao-Yu Hsu and Yu-Chiang Frank Wang, published at NeurIPS 2022. The Semantic-Prototype Variational Transformer takes a partial point cloud and its semantic labels and completes the object, and it allows manipulation of different semantic parts.

  3. 2024

    GSNeRF at CVPR 2024

    GSNeRF, accepted at CVPR 2024, lists him as second author after Zi-Ting Chou, with I-Jieh Liu and Yu-Chiang Frank Wang. The Generalizable Semantic Neural Radiance Field produces both novel-view images and the associated semantic maps for unseen scenes.

  4. 2024

    TPA3D text-to-3D generation

    TPA3D, an ECCV 2024 paper with Bin-Shih Wu, Hong-En Chen and Yu-Chiang Frank Wang, is a GAN-based model for fast text-to-3D generation. Trained with 3D shape data and their rendered 2D images, it uses attention over sentence-level and word-level text features to synthesize 3D mesh data.

  5. 2025

    3DGIC inpainting at CVPR 2025

    In 2025, he is first author of 3D Gaussian Inpainting with Depth-Guided Cross-View Consistency, a CVPR paper written with Zi-Ting Chou and Yu-Chiang Frank Wang. The 3DGIC framework uses rendered depth from each training view to find background pixels visible across views, updates the inpainting mask, and refines the 3D Gaussian Splatting model for inpainting.

  6. 2026

    PhD from National Taiwan University

    He completes his Ph.D. at National Taiwan University in 2026, advised by Prof. Yu-Chiang Frank Wang in the Vision and Learning Lab.

  7. 2026

    OpenVoxel at CVPR 2026

    In 2026, he is first author of OpenVoxel, a CVPR 2026 paper written with Jaesung Choe, Yu-Chiang Frank Wang and Cheng Sun. The training-free algorithm groups and captions sparse voxels from a sparse voxel rasterization model, using Vision Language Models and Multi-modal Large Language Models to build a scene map for open-vocabulary segmentation and referring expression segmentation.

  8. As of October 2026

    Scientist at NVIDIA Research Taiwan

    He is a Research Scientist at NVIDIA Research Taiwan after completing his Ph.D. in 2026. His research topics are mainly on 3D vision, including point cloud perception, point cloud completion, multi-modal 3D generation, 3D scene understanding, and 3D scene reconstruction.

Career

  1. Since 2026NVIDIAResearch Scientist

Education

  1. 2026National Taiwan UniversityPh.D.
  2. 2019National Taiwan UniversityB.S. in Electrical Engineering

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. TPA3D: Triplane Attention for Fast Text-to-3D Generation2024 · Lecture notes in computer science · 3 citationsWith Yu-Chiang Frank Wang
  2. Learning Shape-Color Diffusion Priors for Text-Guided 3D Object Generation2025 · IEEE Transactions on Multimedia · 1 citationsWith Chi-Pin Huang and Yu-Chiang Frank Wang

From OpenAlex, as of 21 October 2026.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Sheng-Yu Huang. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  2. Google has the highest combined h-index among AI organizationsIncludes NVIDIA. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes NVIDIA. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes NVIDIA. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Sheng-Yu Huang yet. Send us a free-licensed photo.

  1. Sheng-Yu Huang | NVIDIA Research Taiwan research.nvidia.com. Checked 20 October 2026.
  2. SPoVT: Semantic-Prototype Variational Transformer for Dense Point Cloud Semantic Completion | NVIDIA Research Taiwan research.nvidia.com. Checked 20 October 2026.
  3. GSNeRF: Generalizable Semantic Neural Radiance Fields with Enhanced 3D Scene Understanding (arXiv:2403.03608) arxiv.org. Checked 20 October 2026.
  4. TPA3D: Triplane Attention for Fast Text-to-3D Generation (arXiv:2312.02647) arxiv.org. Checked 20 October 2026.
  5. 3D Gaussian Inpainting with Depth-Guided Cross-View Consistency (CVPR 2025 Open Access Repository) openaccess.thecvf.com. Checked 20 October 2026.
  6. OpenVoxel: Training-Free Grouping and Captioning Voxels for Open-Vocabulary 3D Scene Understanding (arXiv:2601.09575) arxiv.org. Checked 20 October 2026.
  7. OpenVoxel: Training-Free Grouping and Captioning Voxels for Open-Vocabulary 3D Scene Understanding | NVIDIA Research Taiwan research.nvidia.com. Checked 20 October 2026.