I am a graduating PhD candidate in computer graphics at Washington University in St. Louis, under the supervision of Prof. Tao Ju. During my doctoral work, I focused on procedural geometry analysis for various applications, including maize root systems and cancer tissue. I have made innovative contributions to computer graphics modeling, specifically by devising methods to generate more robust geometric models represented by multiple implicit functions or neural networks. Currently, I am collaborating with Adobe Research to develop an innovative method for generating more precise swept volumes to address challenges in robotics, autonomous driving, and AR/VR drawing.
I graduated from Washington University in Saint Louis in 2021 with double Majors in Computer Science and Mathematics.
📝 Publications

Adaptive Grid Generation for Discretizing Implicit Complexes
Yiwen Ju, Xingyi Du, Qingnan Zhou, Nathan Carr, Tao Ju
-
The method generates a simplicial grid for adaptive discretization of implicit complexes, extending beyond implicit surfaces to represent non-smooth, non-manifold structures.
-
Unlike existing techniques, it adapts to both surfaces and their lower-dimensional intersections, improving accuracy and detail preservation.
-
It supports implicit surface arrangements, CSG shapes, material interfaces, and curve networks, enhancing versatility in geometric modeling.
-
The approach enables efficient, high-fidelity discretization for applications in graphics, material science, and simulations.

TopoRoot+: Computing Whorl and Soil Line Traits of Maize Roots from CT Imaging
Yiwen Ju, Alexander E. Liu, Kenan Oestreich, Christopher N. Topp, Tao Ju
-
Performs Fine-grained root trait extraction based on TopoRoot and provides detailed architectural traits (e.g., root number, length, thickness, angle, hierarchy) from 3D maize root CT images, addressing gaps in existing methods.
-
Outperforms existing methods on real and simulated root images, improving precision in trait measurement.
-
Demonstrated utility in differentiating root mutants from wild types, aiding genetic studies of root architecture and crop productivity.

TopoRoot: a method for computing hierarchy and fine-grained traits of maize roots from 3D imaging
Dan Zeng, Mao Li, Ni Jiang, Yiwen Ju, Hannah Schreiber, Erin Chambers, David Letscher, Tao Ju & Christopher N. Topp
-
Computes fine-grained root traits (e.g., number, length, thickness, angle, tortuosity, branching) from 3D images.
-
Uses topological simplification, geometric skeletonization, and customized heuristics to improve accuracy.
-
Validated on real and simulated maize root images, outperforming existing methods.
-
Runs autonomously within minutes on a desktop with minimal human input.
-
Enables large-scale phenomic studies for genetic research and crop productivity improvements.
📖 Educations
- 2021.01 - Now, Ph.D. in Computer Science, Washington University in Saint Louis
- 2017.09 - 2021.01, B.E. in Computer Science, Washington University in Saint Louis
- 2017.09 - 2021.01 Major in Mathematics, Washington University in Saint Louis
💬 Invited Talks
-
2024.08, Implicit Surface Networks nTop -
2024.07, Adaptive Grid Generation for Discretizing Implicit Complexes Siggraph