I am Yangle Liu, an undergraduate student at the University of Liverpool. My research spans computer vision/graphics and machine learning, with interests in LLMs, real-time scene understanding and editing, high-fidelity rendering, and data-driven methods for robust visual computing.
I am joint supervised by Prof. Jieming Ma (Xi’an Jiaotong–Liverpool University) and Prof. Dominik Wojtczak (University of Liverpool). Additionally, I have engaged in extensive collaborative research with international scholars such as Prof. Hai-Ning Liang (HKUST–Guangzhou), Prof. Yaochun Shen (University of Liverpool), and Postdoc Bo Xiong (Stanford University), all of whom remain my close academic collaborators.
I have interned as a Machine Learning Intern at Neurova AI (New York) and a Software Development Intern at AUO Digitech (Suzhou).
I’m happy to collaborate with researchers from any field, feel free to contact me via email.
🔥 News
- 2026.01: My first-author paper was accepted by ICASSP 2026! 🎉🎉
- 2025.08: My collaborative paper will be presented as an Oral Presentation at BDAI, 2025! 🎉🎉
- 2024.11: The first collaborative paper I worked on was accepted by ICSTIS, 2024!!!
📝 Publications
Endo-G²T: Geometry-Guided & Temporally Aware Time-Embedded 4DGS for Endoscopic Scenes
Yangle Liu, Fengze Li, Kan Liu, Jieming Ma*
- We propose ENDO-G²T, a geometry-guided and temporally aware training scheme for time-embedded 4D Gaussian splatting, achieving state-of-the-art reconstruction accuracy and efficiency in dynamic endoscopic scenes.
CountingFruit: Real-Time 3D Fruit Counting with Language-Guided Semantic Gaussian Splatting
Fengze Li, Yangle Liu, Jieming Ma*, Hai-Ning Liang, Yaochun Shen, Huangxiang Li, Zhijing Wu
- We propose FruitLangGS, a language-guided 3D fruit counting framework that uses adaptive Gaussian splatting, semantic embedding, and prompt-based filtering to efficiently count fruit instances in orchard environments, even under heavy occlusion.
SEED: A Structural Encoder for Embedding-Driven Decoding in Time Series Prediction with LLMs
(Oral Presentation)
Fengze Li, Yue Wang, Yangle Liu, Dou Hong, Jieming Ma*, Huangxiang Li
- We propose SEED, a structural encoder for embedding-driven decoding, designed to bridge the gap between structural representation learning and semantic inference in multivariate time series forecasting, enabling effective integration with pretrained language models for improved prediction accuracy.
EAST: Environment-Aware Stylized Transition Along the Reality-Virtuality Continuum
Under Review
Xiaohan Zhang, Kan Liu, Yangle Liu, Fangze Li, Jieming Ma and Yue Li*
- We propose EAST (Environment-Aware Stylized Transition), a framework that uses 3D Gaussian Splatting and style transfer techniques to seamlessly integrate real-world interruptions into virtual environments, ensuring aesthetic consistency and minimizing disruption along the reality-virtuality continuum.
Automated Generation of Parking Data Sets for Underground Car Parks
Jiakai Li, Yangle Liu, Zheng Rong
- We propose a method using Blender software to generate synthetic underground parking datasets for autonomous driving, addressing challenges in data acquisition and annotation, and providing open-source resources for further development.
🎖 Honors and Awards
- 2023 RoboMaster Super Competition: First Prize (Regional), Second Prize (National)
- 2023 College Student Information System Innovation Competition: Outstanding Award
- 2023 Campus Technology Innovation Association: Managed 3D Modeling Project
- 2022 Campus Ambassador for XJTLU: Achieved promotional targets with six schools, received Pioneer Award
📖 Educations
- 2024.09 - now, Undergraduate, University of Liverpool, UK.
- 2022.09 - 2024.06, Undergraduate, Xi’an Jiaotong–Liverpool University, Suzhou.
💻 Internships
Neurova, New York
Machine Learning Intern
2025.06 - 2025.08
.png)
AUO, Suzhou
Software Development Intern, NUVA Platform Development Department
2024.06 - 2024.08