Qihao Liu
Hello! I am a third-year Ph.D. student in Computer Science (CS) at Johns Hopkins University (JHU), advised by Prof. Alan Yuille. I hold a Master of Robotics at JHU. My research interests lie in the fields of 3D computer vision, generative models, robustness, and robotics.
Before that, I received my Bachelors from Shanghai Jiao Tong University. I have also spent great time at ByteDance (with Song Bai, Liang-Chieh Chen).
Research Interests
My research interests lie in the fields of 3D computer vision, generative models, robustness, and robotics. Specifically, my recent work focuses on:
- Image/3D generation and manipulation [ICLR24, CVPR24, NeurIPS24]
- Evaluating and improving ML system with adversarial examination [CVPR23, ICLR24]
- Video instance segmentation and tracking [ECCV22, CVPR23]
- Human/articulated object pose estimation [Arxiv20, ECCV22]
- Reinforcement learning for robot manipulation [IROS21]
News
- [2024.09] One first-author paper is accepted at NeurIPS 2024.
- [2024.02] Two papers (including one first-author paper) are accepted at CVPR 2024.
- [2024.01] Two first-author papers are accepted at ICLR 2024, including one spotlight.
- [2023.07] One papers is accepted at ICCV 2023.
- [2023.02] Two first-author papers are accepted at CVPR 2023.
- [2022.07] Two first-author papers are accepted at ECCV 2022, including one oral.
- [2022.03] One paper is accepted at CVPR 2022 for oral presentation.
- [2021.07] One first-author paper is accepted at IROS 2021.
- [2021.04] I will join the Computational Cognition, Vision, and Learning (CCVL) Lab as a Ph.D. student at Johns Hopkins University in August 2021.
Publications
Selected Papers:
Alleviating Distortion in Image Generation via Multi-Resolution Diffusion Models
Qihao Liu*, Zhanpeng Zeng*, Ju He*, Qihang Yu, Xiaohui Shen, Liang-Chieh Chen
Neural Information Processing Systems (NeurIPS), 2024
[TL;DR] DiMR is a new diffusion backbone that achieves state-of-the-art image generation. For example, on the ImageNet 256 x 256 benchmark, DiMR with only 505M parameters surpasses all existing image generation models of various sizes, without any bells and whistles.
( arxiv | project page | code )
Qihao Liu*, Zhanpeng Zeng*, Ju He*, Qihang Yu, Xiaohui Shen, Liang-Chieh Chen
Neural Information Processing Systems (NeurIPS), 2024
[TL;DR] DiMR is a new diffusion backbone that achieves state-of-the-art image generation. For example, on the ImageNet 256 x 256 benchmark, DiMR with only 505M parameters surpasses all existing image generation models of various sizes, without any bells and whistles.
( arxiv | project page | code )
DIRECT-3D: Learning Direct Text-to-3D Generation on Massive Noisy 3D Data
Qihao Liu, Yi Zhang, Song Bai, Adam Kortylewski, Alan Yuille
Computer Vision and Pattern Recognition Conference (CVPR), 2024
[TL;DR] DIRECT-3D is a new text-to-3D generative model that directly generates 3D contents in a single forward pass without optimization. It also provides accurate and effective 3D geometry prior for other tasks.
( arxiv | project page | code )
Qihao Liu, Yi Zhang, Song Bai, Adam Kortylewski, Alan Yuille
Computer Vision and Pattern Recognition Conference (CVPR), 2024
[TL;DR] DIRECT-3D is a new text-to-3D generative model that directly generates 3D contents in a single forward pass without optimization. It also provides accurate and effective 3D geometry prior for other tasks.
( arxiv | project page | code )
Generating Images with 3D Annotations using Diffusion Models
Wufei Ma*, Qihao Liu*, Jiahao Wang*, Xiaoding Yuan, Angtian Wang, Yi Zhang, Zihao Xiao, Guofeng Zhang, Beijia Lu, Ruxiao Duan, Yongrui Qi, Adam Kortylewski, Yaoyao Liu, Alan Yuille
(*Equal contribution)
International Conference on Learning Representations (ICLR), Spotlight (Top5%), 2024
( arxiv | project page | code | dataset )
Wufei Ma*, Qihao Liu*, Jiahao Wang*, Xiaoding Yuan, Angtian Wang, Yi Zhang, Zihao Xiao, Guofeng Zhang, Beijia Lu, Ruxiao Duan, Yongrui Qi, Adam Kortylewski, Yaoyao Liu, Alan Yuille
(*Equal contribution)
International Conference on Learning Representations (ICLR), Spotlight (Top5%), 2024
( arxiv | project page | code | dataset )
Discovering Failure Modes of Text-guided Diffusion Models via Adversarial Search
Qihao Liu, Adam Kortylewski, Yutong Bai, Song Bai, Alan Yuille
International Conference on Learning Representations (ICLR), 2024
( arxiv | project page )
Qihao Liu, Adam Kortylewski, Yutong Bai, Song Bai, Alan Yuille
International Conference on Learning Representations (ICLR), 2024
( arxiv | project page )
In Defense of Online Models for Video Instance Segmentation
Junfeng Wu*, Qihao Liu*, Yi Jiang, Song Bai, Alan Yuille, Xiang Bai
(*Equal contribution)
European Conference on Computer Vision (ECCV), Oral, 2022
( PDF | code (☆ 585) | 1-st place solution for CVPR2022 VIS workshop )
Junfeng Wu*, Qihao Liu*, Yi Jiang, Song Bai, Alan Yuille, Xiang Bai
(*Equal contribution)
European Conference on Computer Vision (ECCV), Oral, 2022
( PDF | code (☆ 585) | 1-st place solution for CVPR2022 VIS workshop )
Nothing but Geometric Constraints: A Model-free Method for Articulated Object Pose Estimation
Qihao Liu, Weichao Qiu, Weiyao Wang, Gregory Hager, Alan Yuille
( arxiv )
Qihao Liu, Weichao Qiu, Weiyao Wang, Gregory Hager, Alan Yuille
( arxiv )
Learning Part Segmentation through Unsupervised Domain Adaptation from Synthetic Vehicles
Qing Liu, Adam Kortylewski, Zhishuai Zhang, Zizhang Li, Mengqi Guo, Qihao Liu, Xiaoding Yuan, Jiteng Mu, Weichao Qiu, Alan Yuille
Computer Vision and Pattern Recognition Conference (CVPR), Oral, 2022
( PDF | dataset )
Qing Liu, Adam Kortylewski, Zhishuai Zhang, Zizhang Li, Mengqi Guo, Qihao Liu, Xiaoding Yuan, Jiteng Mu, Weichao Qiu, Alan Yuille
Computer Vision and Pattern Recognition Conference (CVPR), Oral, 2022
( PDF | dataset )