image

My name is Peng Dai. I am broadly interested in Data Mining and Machine Learning, specifically in areas of Generative AI, Deep Learning, Computer Vision, Bioinformatics, and Audio Signal Processing. In particular, my projects cover the practical applications of Large Language Models (LLM) in 3D.

Google Scholar, LinkedIn, ResearchGate

We're hiring PhD fresh graduates, Postdocs, and interns! Feel free to contact me if you're interested.

Currently, I'm working as Principal Scientist with Huawei Canada Research Center, Noah's Ark Lab, working on deep learning with special focus on computer vision applications. I'm Senior Member of IEEE and Senior Member of ACM. Since Nov 2021, I'm the Chair of IEEE Computer Society, Toronto Chapter. From 2018 to 2020, I was an Adjunct Professor at Department of Electrical, Computer & Biomedical Engineering, Ryerson University. I'm in close collaboration with Prof Huijun Ding at Shenzhen University.

My Awesome Lists

  • Awesome 3D Gaussian.
  • Awesome Dynamic NeRF/3DGS.
  • Awesome LLMs.
  • Top News (complete list)

  • [Research] 7/2024, 2 papers accepted at ECCV 2024.
  • [IEEE] 2/2024, we're organizing Workshop on 3D Multimedia Analytics, Search and Generation at ICME 2024.
  • [Product] 9/2023, Our work in Neural Rendering is featured in Huawei Connect 2023.
  • [Research] 9/2023, 1 paper, Decorate3D: Text-Driven High-Quality Texture Generation for Mesh Decoration in the Wild, accepted at NeurIPS 2023.
  • [Research] 8/2023, 1 paper, HiVLP: Hierarchical Interactive Video-Language Pre-Training, accepted at ICCV 2023.
  • [Research] 3/2023, 1 paper accepted at CVPR2023.
  • [Research] 7/2022, 1 paper accepted at ECCV2022.
  • [Research] 6/2022, our team win the 2nd place in 2022 ActivityNet Temporal Action Localization Challenge.
  • [Research] 2/2022, Our paper, "Decompose the Sounds and Pixels, Recompose the Events", has been chosen as Oral presentation at AAAI2022 (Feb 26, 10:30am-11:45am PST)! Also featured in Noah's Ark official blog! [Project Page]
  • [Research] 2/2022, Noah's Ark Lab has open sourced a large-scale multi-modality chinese dataset Wukong. arxiv.
  • [Research] 12/2021, two papers accepted by AAAI2022. [CPNet] [EDRNet]
  • [IEEE] 12/2021, IEEE Engineering and Human Environment joint Chapter is organizing a Talk.
  • [Product] 11/2021, Huawei Intelligent Education Solution in the News (Chinese).
  • [Product] 10/2021, Our collaboration team developed Sign Language Digital Avatar.
  • [Commercial] 10/2021, we become the Spotlight sponsor of ICCV2021.
  • [Research] 07/2021, two papers accepted by ICCV2021: CSA Video, [Project Page]
  • [Product] 07/2021, Petal Search for Sports Highlights, News.
  • [Product] 07/2021, Check the new feature in HUAWEI Watch 3, News.
  • [Research] 06/2021, our team win the 2nd place in 2021 ActivityNet challenge, presentation video, [Project Page].
  • [Product] 2020, Video Labeling system for Video Content Tagging (VCT) URL.
  • [Research] 07/2020, two papers accepted by ECCV2020.
  • [Product] 2018, Spotlight Reel is featured in the Huawei Release conference!! Press, Commercial, News
  • Project Hightlights

    Picture

    Spotlight Reel, released in 2018
    Semi-supervised hybrid clustering/classification system, Patent No.: US11023710B2.
    Intelligent video interaction method, Patent No.: US10979761B2.
    [News] [Press] [Commercial] [Commercial]


    Mate 20 series mobile phone's unique identity-oriented AI video cutting function can extract the key segment of the specific main character. With various special effects, these clips are automatically compiled into wonderful movies about the character. Gather the bits you want and build your own blockbuster.

    Picture

    ActivityNet Challenge, 2021
    Class Semantics-based Attention for Action Detection, ICCV2021.
    Towards Efficient Coarse-to-Fine Networks for Action and Gesture Recognition, ECCV2020.
    [CSA ] [C2F] [Project Page]

    We won the 2nd place in 2021 ActivityNet Temporal Action Localization Challenge.

    Picture

    Sports Highlights for Petal Search, released in 2021
    Patent filed
    [News] [News (Tokyo Olympics)] [C2F] [CPNet]

    Open Petal Search and search for your favorite athletes and sports, and you'll find the latest news, event times, medal counts, athlete information, video highlights, and a wealth of other intriguing content. Whether you're interested in the stats and scores, or are following your favorite athletes, with Petal Search, there's no need to miss a moment of sports action!

    Past Experience

    From Jan 2015 to Dec 2016, I was with the Department of Computer Science, University of Western Ontario, also affiliated with Robarts Research Institute. I worked on the modeling and detection of Alzheimer's Disease (AD) with Dr. Femida Gwadry-Sridhar and Prof. Michael Bauer as my Principal Investigators. I also worked closely with Prof. Michael Borrie.

    From Jan 2014 to Dec 2014, I worked as a Postdoctoral Fellow in Toronto Rehabilitation Institute (TRI), joint affiliated with the Department of Computer Science, University of Toronto. TRI is a member of University Health Network (UHN) and is affiliated with the University of Toronto. University Health Network (UHN) is Canada's largest research hospital, and a major teaching hospital of the University of Toronto. My supervisors were Assit/P Frank Rudzicz (Department of Computer Science, University of Toronto) and A/P Alex Mihailidis (Dept of Occupational Science and Occupational Therapy & Institute of Biomaterials and Biomedical Engineering, University of Toronto).

    From July 2012 to Dec 2013, I worked as a Research Associate for Assist/P Ho Shen-Shyang in School of Computer Engineering (SCE) of Nanyang Technological University (NTU). I was with the Center for Computational Intelligence (C2i).