Ziyi Wu (吴紫屹)

  Ph.D. Student

  University of Toronto

  Email: ziyiwu [at] cs.toronto.edu

  CVGoogle ScholarGitHubTwitter

About

I am a third-year PhD student in Toronto Intelligent Systems Lab (TISL) at the University of Toronto, supervised by Prof. Igor Gilitschenski. I also interned at Google Research working with Dr. Thomas Kipf. I had the fortune to work with Prof. Animesh Garg, Prof. Andrea Tagliasacchi. I am affiliated with the Vector Institute.

Prior to my Ph.D., I received my Bachelor's degree from the Department of Automation, Tsinghua University, where I worked with Prof. Jiwen Lu. I spent an unforgettable summer in 2020 doing internship at Stanford University, advised by Prof. Leonidas J. Guibas.

I am broadly interested in designing machine learning systems that can perceive the world as we humans do, which I believe is the cornerstone of achieving AGI. I have been working on object-centric learning (structured representation), event-based vision (silicon retina), and generative AI (world models).


News


Research

* indicates equal contribution/supervision


Preprints

SPAD

SPAD : Spatially Aware Multiview Diffusers
Yash Kant, Ziyi Wu, Michael Vasilkovsky, Guocheng Qian, Jian Ren, Riza Alp Guler,
Bernard Ghanem, Sergey Tulyakov*, Igor Gilitschenski*, Aliaksandr Siarohin*
(under review)
[Project Page]

S2Edit

S2Edit: Text-Guided Image Editing with Precise Semantic and Spatial Control
Xudong Liu, Zikun Chen, Ruowei Jiang, Ziyi Wu, Kejia Yin, Han Zhao,
Parham Aarabi, Igor Gilitschenski
(under review)

LEOD

LEOD: Label-Efficient Object Detection for Event Cameras
Ziyi Wu, Mathias Gehrig, Qing Lyu, Xudong Liu, Igor Gilitschenski
(under review)
[Paper]


Publications

SlotDiffusion

SlotDiffusion: Object-Centric Generative Modeling with Diffusion Models
Ziyi Wu, Jingyu Hu*, Wuyue Lu*, Igor Gilitschenski, Animesh Garg
NeurIPS 2023 | ICLR@NeSy-GeMs workshop, 2023.
Spotlight Presentation
[Project Page (with paper and code)]  [Slides]

SlotFormer

SlotFormer: Unsupervised Visual Dynamics Simulation with Object-Centric Models
Ziyi Wu, Nikita Dvornik, Klaus Greff, Thomas Kipf*, Animesh Garg*
ICLR 2023 | UAI@CRL workshop, 2022 | ECCV@MVCS Challenge, 2022.
[Project Page (with paper and code)]  [Slides]

Breaking Bad

Breaking Bad: A Dataset for Geometric Fracture and Reassembly
Silvia Sellán*, Yun-Chun Chen*, Ziyi Wu*, Animesh Garg, Alec Jacobson
NeurIPS 2022 Datasets and Benchmarks Track.
Featured Paper Presentation
[Project Page (with paper, code and data)]  [Slides]

ISL

Instance Similarity Learning for Unsupervised Feature Representation
Ziwei Wang, Yunsong Wang, Ziyi Wu, Jiwen Lu, Jie Zhou
ICCV 2021.
[Paper]  [Code]

AutoBiDet

Learning Efficient Binarized Object Detectors with Information Compression
Ziwei Wang, Jiwen Lu, Ziyi Wu, Jie Zhou
T-PAMI 2021.
[Paper]

BiDet

BiDet: An Efficient Binarized Object Detector
Ziwei Wang, Ziyi Wu, Jiwen Lu, Jie Zhou
CVPR 2020.
[Paper]  [Code]  [Slides]


Academic Services

Journal reviewer: T-PAMI; ISRR;
Conference reviewer: CVPR'22-24; ICCV'23; ECCV'22-24; NeurIPS'22-23; ICML'23-24; ICLR'24; AAAI'23-24; IJCAI'24; IROS'22; ICRA'23-24;
Workshop reviewer: OSC@ICLR'22;



Teaching

Teaching Assistant, CSC 478: Robotics Perception23 Winter
Teaching Assistant, CSC 108: Introduction to Computer Programming21 Fall, 22 Winter, 22 Fall, 23 Summer



Selected Awards


Miscellaneous

I like basketball and soccer. My favorite players are LeBron James and Lionel Messi (G.O.A.T.!).

My favorite books are One Hundred Years of Solitude, 1984, The Plague, and Dream of the Red Chamber (红楼梦). My favorite writers are García Márquez and Albert Camus. Some of my recent (since 2017) favorites are Hermann Hesse, Milan Kundera, Mo Yan (莫言), Tocqueville, Yukio Mishima, Yasunari Kawabata, Proust, etc. My main interests are ethics, human nature and values.

I am also an Anime fan and a lover of Visual Novel, and JRPG (added, 2023.1). Some of my favorite works are Fate/Stay Night, EVA, Perfect Blue, Hyouka, Subarashiki Hibi (a.k.a. Wonderful Everyday), and Persona 5 (added, 2023.1).

Two persons I admire the most: Kaiming He for his outstanding research tastes ("All great truths are simple in final analysis") and Elon Musk for his aspiration and enthusiasm (now in doubt, 2022.12. "Power tends to corrupt and absolute power corrupts absolutely").

I write an article (in Chinese) summarizing the experience during my CS PhD application process. Take a look here!




Ziyi Wu Last updated: Feb.12, 2024