Description of the image
Crane He Chen

Ph.D. candidate in Computer Graphics
half-time researcher @ Adobe / half-time teaching faculty @ Hopkins

I am a graduate student researcher specializing in surface reconstruction and geometric data processing. I am pursuing a Ph.D. degree in engineering at The Johns Hopkins University under the guidance of prof. Misha Kazhdan and prof. Noah Cowan. I worked concurrently towards Computer Science and Mechanical Engineering degrees at Hopkins, which means I accomplished enough courseworks from both departments.

I am an enthusiast of images, surfaces, and volumes. I love my research. Computer Graphics unites the disciplines of mathematics, physics, and art, creating a captivating field of study. The dynamic nature of this intersection presents a plethora of exhilarating opportunities for exploration and discovery. I cannot think of more exciting things to investigate than this.

I trained myself in geometry for the second half of my PhD. Before that, I trained myself in 3d vision/ML. In addition, I spent a year working at Apple during the PhD, in the team led by Joerg Liebelt. All the trainings have enabled me to effectively observe and address intriguing real-world challenges utilizing my geometric knowledge.

I also devoted time to reading papers on diversity and inclusion, advocating for a welcoming work atmosphere that supports underrepresented communities in computer science and engineering. I served the position of social chair at GRACE and offered mentorship through WiCS at Hopkins.


I enjoy coding, and actually, if I go too many consecutive days without it, I start to feel a bit strange.

Image description

• A C++ total curvature calculator works universally for both triangle meshes and point clouds, that is fast and robust.

• Potenetial upcoming new feature for popular libraries, submitting to Libigl and Open3D.

Image description

• A Swift api for arkit, released at WWDC 2022.

• I had the fortunate opportunity to be involved in this project during my time at Apple.

Image description

• An extensively streamlined Matlab implementation of PoissonRecon.

• Friendly for beginners, no octree, no any blackbox, implemented multigrid solver from scratch.


Image description
Estimating Discrete Total Curvature with Per Triangle Normal Variation


Crane Chen

(project supervised by Misha Kazhdan, funded by departmental fellowship)

Image description
Towards Undoing Smoothness Bias for RGB Neural Surface Reconstruction

Under Review, 2023

Crane Chen, Joerg Liebelt

(project supervised by Misha Kazhdan, funded by Apple Inc.)

Image description
Towards Efficient Graph Convolutional Networks for Point Cloud Handling

ICCV, 2021

Yawei Li*, Crane Chen*, Zhaopeng Cui, Radu Timofte, Marc Pollefeys, Greg Chirikjian, Luc Van Gool

* stands for equal contributors

Image description
Multi-person 3D Pose Estimation in Crowded Scenes Based on Multi-View Geometry


Crane Chen*, Pengfei Guo*, Pengfei Li, Gim Hee Lee, Greg Chirikjian

* stands for equal contributors

Side Projects

Apart from research, I have a broader interest in problem-solving.

At Hopkins, it is typical for students to be assigned real-world issues as group projects. I spent a greater amount of time than most on enjoyable side projects.

Augmented Reality Suggestive Contours for Brain Surgerydemo

• Developed two apps for Microsoft Hololens with Unity, which involved features including gesture control, voice control, and gaze tracking.

• Calculated suggestive contour of CAD models using OpenGL libraries. Applied the Vuforia marker to detect the 3D model position and overlay suggestive contours on a 3D printed model.

Occlusion R-CNN for Pedestrian Detectiondemo

• Appeared on the news at JHU.

• Provided a solution for occlusion situation in pedestrian detection.

UR5 Robot Arm Playing Pianodemo

• Developed an app with MATLAB and ROS, which enables UR5 automatically read-understand-play any encoded music score (i.e. a data set encoding notes).

Plain Academic