Chengyang Zhao

I am a final-year undergraduate student at Yuanpei College, Peking University in China, majoring in Data Science and Big Data Technology (Statistics + Computer Science). I am fortunate to be advised by Prof. He Wang. I am also privileged to work closely with Prof. Chuang Gan and Prof. Yunzhu Li.

My research interest is broadly in computer vision (especially 3D vision), robotics, and multi-modal learning. My research objective is to build an intelligent agent with comprehensive perception, reasoning, and execution capabilities developed from multi-modal information, which can interact effectively and efficiently with the real world.

Email / Twitter / Github

Research

	GAPartNet: Cross-Category Domain-Generalizable Object Perception and Manipulation via Generalizable and Actionable Parts Haoran Geng, Helin Xu, Chengyang Zhao, Chao Xu, Li Yi, Siyuan Huang, He Wang ( Equal contribution with the order determined by rolling dice.) CVPR, 2023, Highlight (top 2.5% of all submissions) with all top ratings [paper] [webpage] [code] [dataset] We propose to learn cross-category generalizable object perception and manipulation skills via Generalizable and Actionable Parts (GAParts). We present GAPartNet, a large-scale interactive dataset with rich, part-level annotations for both perception and interaction tasks.
	TextPSG: Panoptic Scene Graph Generation from Textual Descriptions Chengyang Zhao, Yikang Shen, Zhenfang Chen, Mingyu Ding, Chuang Gan ICCV, 2023 [paper] [webpage] [code] We introduce a novel problem aiming to learn panoptic scene graph generation entirely from textual descriptions. We design a modularized proposal-free framework, which not only breaks the generalization limitation within previous detector-based methods but also learns extensive and various object semantics and relation predicates from text.
	Controllable 3D Scene Editing with Sparse Neural Radiance Fields Mingtong Zhang, Chengyang Zhao, Yining Hong, Hongsheng Lu, Chuang Gan (* Equal contribution.) CVPR, 2024, In Submission We propose a novel scene representation Sparse Neural Radiance Fields to compartmentalize the 3D scene into specialized expert fields, and design a DINO-based gating mechanism for automatic semantics-based scene decomposition across the expert fields during reconstruction. Our proposed representation effectively enhances controllability, enabling more precise and intuitive editing on 3D scenes.

Experience

	University of Illinois Urbana-Champaign 2023.06 - Present Undergraduate Research Intern (Remote) Research Advisor: Prof. Yunzhu Li
	Massachusetts Institute of Technology 2022.08 - 2023.05 Visiting Undergraduate Student Research Advisor: Dr. Chuang Gan
	Peking University 2021.05 - Present Undergraduate Research Intern Research Advisor: Prof. He Wang
	Peking University 2019.09 - Present Undergraduate Student, Yuanpei College GPA ranking: 2/13

Selected Awards and Honors

• SenseTime Scholarship (30 undergraduate students/year in China), SenseTime, 2023

• Third Prize of Peking University Scholarship, Peking University, 2022

• Award for Academic Excellence, Peking University, 2022

• Xiaomi Scholarship, Peking University, 2020

• Merit Student, Peking University, 2020

Thanks Jon Barron for this amazing template :D
Last Updated: Dec. 30, 2023