Stephanie Wenxin Liu

I am a Research Scientist in the Image Synthesis and Machine Learning group at The Max Planck Institute for Informatics, collaborating with Dr. Thomas Leimkühler. My research focuses on the intersection of computer graphics, computer vision and machine learning.

Recently, I completed my MSc Computer Science at Birkbeck, University of London, collaborating with Prof. Tobias Ritschel (University College London, UCL) on the intersection of computer graphics and machine learning, in particular, applying neural networks to bounding volumes and spatial queries. Our article, “Neural Bounding”, is published at SIGGRAPH 2024.

Alongside my academic pursuits, I am a senior engineer with a decade of industry experience. My professional expertise spans across data, infrastructure, platform, and software. In particular, I specialised extensively in Site Reliability Engineering (distributed systems, networking, platforms, pipelines, telemetry, Linux, cloud computing) and Software Engineering (object-oriented: Go, Java, Python; Functional: Scala, Kotlin.)

I received my BSc (Hons) from University College London. Looking ahead, I’m interested in pursuing further research via a Ph.D. that combines computer graphics and computer vision.

You can get in touch with me at wenxin.liu.cs[at]gmail.com.

Latest publication

Neural Bounding

Stephanie Wenxin Liu, Michael Fischer, Paul D. Yoo, and Tobias Ritschel

Proceedings of the Special Interest Group on Computer Graphics and Interactive Techniques Conference Papers ’24 (SIGGRAPH Conference Papers ’24), 2024

Abs HTML PDF

Bounding volumes are an established concept in computer graphics and vision tasks but have seen little change since their early inception. In this work, we study the use of neural networks as bounding volumes. Our key observation is that bounding, which so far has primarily been considered a problem of computational geometry, can be redefined as a problem of learning to classify space into free or occupied. This learning-based approach is particularly advantageous in high-dimensional spaces, such as animated scenes with complex queries, where neural networks are known to excel. However, unlocking neural bounding requires a twist: allowing – but also limiting – false positives, while ensuring that the number of false negatives is strictly zero. We enable such tight and conservative results using a dynamically-weighted asymmetric loss function. Our results show that our neural bounding produces up to an order of magnitude fewer false positives than traditional methods. In addition, we propose an extension of our bounding method using early exits that accelerates query speeds by 25%. We also demonstrate that our approach is applicable to non-deep learning models that train within seconds. Our project page is at: https://wenxin-liu.github.io/neural_bounding/.