Selected Publications


MonoFusion: Sparse-View 4D Reconstruction via Monocular Fusion
[new updated results] Zihan Wang, Jeff Tan, Tarasha Khurana, Neehar Peri, Deva Ramanan*
In submission. [Paper] [Project] [Code, Data(soon released)]
"The first work addressing free-view synthesis for 4D dynamic scene reconstruction using sparse-view inputs."

AirShot: Efficient Few-Shot Detection for Autonomous Exploration
Zihan Wang, Bowen Li, Chen Wang, Sebastian Scherer*
IROS 2024 [Paper] [Project] [Code]
"Leverage class-agnostic relation to implement few-shot detection w/o fine-tuning, which enables fast and efficient robotic exploration."

Learning with Noisy Foundation Models
Hao Chen, Zihan Wang, Jindong Wang, Ran Tao, Hongxin Wei, Xing Xie, Masashi Sugiyama, Bhiksha Raj
IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)
[Accepted]
[Paper] [Project] [Code(yet to come)]
"First work to comprehensively understand and analyze the nature of noise in pre-training datasets and then effectively mitigate its impacts on downstream tasks."

Identity Preservation for Counterfactual Estimation
Zihan Wang, Bachelor Thesis [Paper] [Project] [Code(yet to come)]
"Novel strategy use cross-section data only, conditioned on disentangled features for identity preservation."

ONLS: Optimal Noise Level Search in Diffusion Autoencoders Without Fine-Tuning
Zihan Wang, International Conference on Learning Representations(ICLR) Workshop 2024
"A simple, effective, and fine-tuning free framework for achieving optimal encoding depth for each sample."

PyPose: A Library for Robot Learning with Physics-based Optimization
Chen Wang, others, Zihan Wang, others
CVPR 2022 & IROS Workshop 2023 [Paper_0] [Paper_1] [Project] [Code] [Video]
"Connecting classic robotics with modern learning methods."

Project

A Style-Preserved Motion Controller for Simulated Humanoid Characters

Robot Learning [Report] [Project] [Code] [Video]
"Proposes a style-preserved motion controller that clusters motion data, trains latent variable models for diverse behaviors, and uses a style-aware high-level controller to enable multi-tasking with varied styles."

PyPose Teaser
PyPose: A Library for Robot Learning with Physics-based Optimization

OpenCV in robotics: Connecting classic robotics with modern learning methods [Paper_0] [Paper_1] [Project] [Code] [Video]
"Contribution to advancing constrained optimization techniques, with applications in trajectory planning and motion control."

MONAI Teaser
MONAI: Medical Open Network for Artificial Intelligence

Medical Open Network for Artificial Intelligence [Project] [Code] [Video]
"Contribution to the development of generative models, particularly focusing on diffusion autoencoders with applications in MRI."

Awards


  • Edinburgh Award (Research Experience) - 2022
  • Edinburgh Scholarship (£4000) - 2022
  • Edinburgh Scholarship (£4000) - 2021
  • QIANJIANG ELECTRIC Scholarship (£3000) - 2020

Service


Academic Reviewer

  • International Conference on Learning Representations (ICLR)
  • European Conference on Computer Vision (ECCV)
  • IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)
  • IEEE Transactions on Image Processing (TIP)
  • IEEE Transactions on Circuits and Systems for Video Technology (TCSVT)

Advisors


Prof. Deva Ramanan@CMU_RI

Prof. Sebastian Scherer@CMU_RI

Prof. Chen Wang@Buffalo

Prof. Sotirios Tsaftaris@UoE

Contact

  • zihanwa3@cs.cmu.edu
  • +1-412-589-8046
  • EDSH, Carnegie Mellon University, Pittsburgh, PA 15213, USA

Visitors

I am confident that my 'boundaryless', 'website-stalking' friends have contributed to at least 74.47% of it.