Pengcheng Yu

dblp:279/0855 · DBLP profile ↗
← Back
4ranked-venue papers
0as first author
3since 2021 · last 2025
—ORCID · conflict

Domains — the database's venue-derived domains; a paper can count in several

Graphics, computer vision, multimedia, augmented reality and games · 2 · 2 since 2021Artificial intelligence and machine learning · 1 · 1 since 2021Systems, architecture and hardware · 1Databases, data management, data science and information retrieval · 1 · 1 since 2021

Expertise — from the expertise taxonomy: the topics of the expert's papers under the CCF categories. A weight counts papers with recency: 1 for a paper about the topic, 0.3 when the topic is its context, halved every five years.

Artificial intelligence
1 paper
3D vision · 75% Video understanding and tracking · 25%

Topics — the 4 heaviest of 4, each with the papers that count most for it

TopicWeightPapersLastEvidence papers
Computer vision › 3D vision › motion estimation › 3d motion tracking
3d point tracking
0.912025
ST4RTrack: Simultaneous 4D Reconstruction and Tracking in the World · ICCV 2025
Computer vision › 3D vision
3d reconstruction
0.912025
ST4RTrack: Simultaneous 4D Reconstruction and Tracking in the World · ICCV 2025
Computer vision › 3D vision › 3d reconstruction
dynamic 3d reconstruction
0.912025
ST4RTrack: Simultaneous 4D Reconstruction and Tracking in the World · ICCV 2025
Computer vision › Video understanding and tracking
feature tracking
0.912025
ST4RTrack: Simultaneous 4D Reconstruction and Tracking in the World · ICCV 2025

Methods — techniques the papers use, named apart from their topics

reprojection loss · 0.9pointmap prediction · 0.9feed-forward framework · 0.9
YearPublicationVenuePosition
2025 ST4RTrack: Simultaneous 4D Reconstruction and Tracking in the World
abstract
Dynamic 3D reconstruction and point tracking in videos are typically treated as separate tasks, despite their deep connection. We propose St4RTrack, a feed-forward framework that simultaneously reconstructs and tracks dynamic video content in a world coordinate frame from RGB inputs. This is achieved by predicting two appropriately defined pointmaps for a pair of frames captured at different moments. Specifically, we predict both pointmaps at the same moment, in the same world, capturing both static and dynamic scene geometry while maintaining 3D correspondences. Chaining these predictions through the video sequence with respect to a reference frame naturally computes long-range correspondences, effectively combining 3D reconstruction with 3D tracking. Unlike prior methods that rely heavily on 4D ground truth supervision, we employ a novel adaptation scheme based on a reprojection loss. We establish a new extensive benchmark for world-frame reconstruction and tracking, demonstrating the effectiveness and efficiency of our unified, data-driven framework. Our code, model, and benchmark will be released.
Haiwen Feng, Junyi Zhang 0004, Qianqian Wang 0002, Yufei Ye 0001, Pengcheng Yu, Michael J. Black, Trevor Darrell, Angjoo Kanazawa
ICCV5
2025 Deep learning-enabled turbulence model optimization of solid motor
Huixin Yang, Pengcheng Yu, Bixuan Lou, Xiang Li 0018
Adv. Eng. Informatics2
2025 Adaptive multi-modal prompting for universal image restoration amidst diverse degradations
Jinhua Wang 0004, Sunhan Xu, Pengcheng Yu, XiaoYue Ma
Vis. Comput.6
2020 An Asymmetrical Loosely Coupled Transformer and Constant Current Wireless Charging Scheme for Warehouse Vehicles
abstract
Wireless power transfer (WPT) is the preferred charging method for battery-powered electric vehicles (EVs). In this paper, a novel WPT system with constant current charging capability is introduced for warehouse vehicles with lead-acid batteries. First, the schematic structure of the proposed system is discussed. A novel structure of asymmetrical loosely coupled transformer is designed to enable high-voltage feeding to low-voltage output and reduce the weight of the secondary side. Secondly, an optimized magnetic coupler using ferrite cores and magnetic shielding structure is proposed to ensure stable power transmission and high efficiency. Then, a dual-side independent control strategy on dual active bridge is proposed, aiming for maintaining constant charging current and high system efficiency. Finally, a 3kW inductive charging experimental platform is established to validate the theoretical analysis. Constant charging current of 45 A is achieved at the transmission distance of 50mm. The rated transferring power and system efficiency are 3 kW and 94.3%, respectively.
Xuecheng Liu, Wenjie Guan, Pengcheng Yu, Pengfei Pan
IECON3