I am a fifth-year Ph.D. student in the i-VisionGroup, Department of Automation, Tsinghua University, advised by Prof. Jianjiang Feng and Prof. Jie Zhou.
I received my B.S. degree in Automation in 2021, together with a minor in the Academy of Arts and Design, both from Tsinghua University.
My current interests and experience lie in the following areas:
One paper on fixed-length dense fingerprint representation was accepted by T-IFS.
One paper on under-screen fingerprint pose estimation was accepted by T-IFS, and one paper on finger pose HCI was accepted by T-MC.
Achieved 2nd place at MER25 @ ACM MM, and had one paper on finger photo pose estimation accepted by IJCB 2025 as Oral.
Achieved 2nd place in CVRR and EgoSchema Challenge at CVPR 2025, with related work accepted by EgoVis @ CVPR 2025.
Doctor of Philosophy (Ph.D.)
Focusing on enhancing performance and efficiency in domain-specific tasks via data & representation normalization / domain-informed modeling / workflow-level optimization; supervised by Prof. Jianjiang Feng and Prof. Jie Zhou.
Resulting in 11 publications in international journals and conferences (6 first-author).
Minor in Digital Media Art
Coursework focusing on digital entertainment content and interactive media, with an emphasis on integrating creative design and interdisciplinary digital technologies.
Bachelor of Engineering (B.Eng.)
Undergraduate studies in automation, covering control theory, artificial intelligence and data analysis, with exposure to system design and engineering applications.
Resulting in 1 publications in international conferences (first-author).
Research Intern
Focusing on multimodal perception & representation, intent understanding, and affective computing, improving deployment efficiency and performance through prompt engineering, post-training, agent collaboration, and pipeline optimization for multimodal large models.
Resulting in 4 technical reports (2 first/corresponding-author), and 4 corresponding awards in international competitions (ACM MM, CVPR).
Algorithm Intern
Focusing on contactless palmprint recognition for large-scale populations, responsible for framework design, technical investigation, and early-stage engineering prototyping. Corresponding project was piloted for palm-based payment on the Beijing Daxing Airport Express line.
Resulting in a Silver Award in Tsinghua University Doctoral Social Practice.
Algorithm Intern (University-Industry Collaboration)
Focusing on image retrieval in edge computing scenarios (partial images / low quaity), improving storage and computational efficiency through image registration and mosaicking technologies.
Resulting in 1 publication in international journal (first-author).
Works on multi-modal large models, post-training, agentic workflow, and harness engineering.
Multi-modal emotion recognition with leveraged signals, samples and deliberation.
Fine-grained multi-modal emotion recognition via multi MLLMs and model ensemble.
Long-form video understanding and robustness evaluation via Chain of Thought and harness engineering.
Works on finger-based human-computer interaction, pose estimation, and sensing.
Leveraging 3D finger pose to improve contactless fingerprint recognition through acquisition guidance and pose constraints.
Works on large-scale image retrieval with efficient representation and geometric normalization.
Fixed-length dense representation and matching framework that couples local discriminability with pose-aware alignment and robustness enhancement.
Rigid pose estimation via multi-modal fusion strategies and its application to retrieval.
CNN-ViT hybrid network for joint pose estimation and identity recognition in partial image scenarios.
Dual-branch dense registration paradigm that integrates precise geometric priors (phase) with robust deep representation.
End-to-end dense distortion field regression replacing previous low-dimensional assumptions with grid-level geometric correction.
End-to-end dense distortion field regression replacing previous low-dimensional assumptions with grid-level geometric correction.