Ziyi Wang

I am a fifth year PhD student in the Department of Automation at Tsinghua University, advised by Prof. Jiwen Lu . In 2020, I obtained my B.Eng. in the Department of Electronic Engineering, Tsinghua University. I also obtained B.Admin. as dual degree in the School of Ecnomics and Management, Tsinghua University.

I am broadly interested in computer vision and deep learning. My current research focuses on 3D vision, 3D generation and 4D world model.

Email / Google Scholar / Github

News

2025-06: 1 survey paper on vision generalist model is accepted to IJCV.

2025-02: 1 paper on unified 3D point cloud pre-training is accepted to CVPR 2025.

2024-09: 1 paper on 3D open vocabulary semantic segmentation is accepted to NeurIPS 2024.

2024-01: The journal paper of P2P is accepted to TPAMI.

2023-07: 1 paper on 3D generative pre-training is accepted to ICCV 2023.

2023-07: The journal paper of PV-RAFT is accepted to TPAMI.

2022-09: 1 paper (spotlight) on 3D prompt learning is accepted to NeurIPS 2022.

2022-03: 1 paper on 3D semantic segmentation is accepted to CVPR 2022.

2021-07: 2 papers (including 1 oral) are accepted to ICCV 2021.

2021-03: 1 paper on 3D scene flow estimation is accepted to CVPR 2021.

Publications

* indicates equal contribution

	Vision Generalist Model: A Survey Ziyi Wang, Yongming Rao, Shuofeng Sun, Xinrun Liu, Yi Wei, Xumin Yu, Zuyan Liu, Yanbo Wang, Hongmin Liu, Jie Zhou , Jiwen Lu International Journal of Computer Vision (IJCV), 2025 [arXiv] We conduct a comprehensive survey on vision generalist models that support multimodal inputs and can handle various downstream tasks.
	OGGSplat: Open Gaussian Growing for Generalizable Reconstruction with Expanded Field-of-View Yanbo Wang, Ziyi Wang*, Wenzhao Zheng, Jie Zhou , Jiwen Lu Preprint*. [arXiv] [Code] [Project Page] OGGSplat is designed to expand the field-of-view of the Gaussian-based 3D scene reconstructed from sparse views and feedforward / generalizable models.
	UniPre3D: Unified Pre-training of 3D Point Cloud Models with Cross-Modal Gaussian Splatting Ziyi Wang, Yanran Zhang, Jie Zhou , Jiwen Lu IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025 [arXiv] [Code] UniPre3D is a unified pre-training method that can be applied to both object-level and scene-level point clouds. It is supported by cross-modal Gaussian splatting technique.
	XMask3D: Cross-modal Mask Reasoning for Open Vocabulary 3D Semantic Segmentation Ziyi Wang, Yanbo Wang, Xumin Yu, Jie Zhou , Jiwen Lu Conference on Neural Information Processing Systems (NeurIPS), 2024 [arXiv] [Code] XMask3D is a framework that propose mask-level reasoning techniques to empower 3D segmentation model with open vocabulary capacity under the assistance of the pre-trained 2D mask generator.
	Point-to-Pixel Prompting for Point Cloud Analysis With Pre-Trained Image Models Ziyi Wang, Yongming Rao, Xumin Yu, Jie Zhou , Jiwen Lu IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2024 [IEEE] [Code] [Project Page] P2P++ is the extended journal version of P2P. We further propose Pixel-to-Point Distillation to make P2P applicable in scene-level perception tasks.
	3D Point-Voxel Correlation Fields for Scene Flow Estimation Ziyi Wang, Yi Wei, Yongming Rao, Jie Zhou , Jiwen Lu IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2023 [IEEE] [Code] [Project Page] DPV-RAFT is the extended journal version of PV-RAFT. We further propose Spatial Deformation and Temporal Deformation to enhance PV-RAFT.
	Take-A-Photo: 3D-to-2D Generative Pre-training of Point Cloud Models Ziyi Wang, Xumin Yu, Yongming Rao, Jie Zhou , Jiwen Lu IEEE International Conference on Computer Vision (ICCV), 2023 [arXiv] [Code] [Project Page] TAP is a 3D-to-2D generative pre-training method that generate projected images of point clouds from instructed perspectives.
	P2P: Tuning Pre-trained Image Models for Point Cloud Analysis with Point-to-Pixel Prompting Ziyi Wang, Xumin Yu, Yongming Rao, Jie Zhou , Jiwen Lu Conference on Neural Information Processing Systems (NeurIPS), 2022 Spotlight* [arXiv] [Code] [Project Page] [中文解读] P2P is a framework to leverage large-scale pre-trained image models for 3D point cloud analysis.
	SemAffiNet: Semantic-Affine Transformation for Point Cloud Segmentation Ziyi Wang, Yongming Rao, Xumin Yu, Jie Zhou , Jiwen Lu IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022 [arXiv] [Code] We present Semantic-Affine Transformation that transforms decoder mid-level features of the encoder-decoder segmentation network with class-specific affine parameters.
	PoinTr: Diverse Point Cloud Completion with Geometry-Aware Transformers Xumin Yu, Yongming Rao, Ziyi Wang, Zuyan Liu, Jiwen Lu , Jie Zhou IEEE International Conference on Computer Vision (ICCV), 2021 Oral Presentation [arXiv] [Code] [中文解读] PoinTr is a transformer-based framework that reformulates point cloud completion as a set-to-set translation problem.
	Towards Interpretable Deep Metric Learning with Structural Matching Wenliang Zhao, Yongming Rao, Zyi Wang, Jiwen Lu , Jie Zhou IEEE International Conference on Computer Vision (ICCV), 2021 [arXiv] [Code] We present a deep interpretable metric learning (DIML) that adopts a structural matching strategy to explicitly aligns the spatial embeddings by computing an optimal matching flow between feature maps of the two images.
	PV-RAFT: Point-Voxel Correlation Fields for Scene Flow Estimation of Point Clouds Yi Wei , Ziyi Wang, Yongming Rao, Jiwen Lu , Jie Zhou IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)*, 2021 [arXiv] [Code] We present point-voxel correlation fields for 3D scene flow estimation which migrates the high performance of RAFT and provides a solution to build structured all-pairs correlation fields for unstructured point clouds.