Chuqiao Li

I am a Ph.D. candidate at the University of Tübingen & Tübingen AI Center, supervised by Prof. Gerard Pons-Moll, Prof. Andreas Geiger, and Prof. Seong Joon Oh. My research focuses on 3D human motion synthesis and understanding, human-scene interaction, and generative AI.

Prior to this, I earned my master’s degree at ETH Zurich, where I developed a strong interest in computer vision, machine learning, digital humans, and scene understanding. For my master’s thesis, I collaborated with Dr. Marc Habermann and Prof. Christian Theobalt at the Visual Computing and Artificial Intelligence department of MPI-INF, focusing on visual SLAM and 3D human pose estimation. During my studies, I also collaborated with Prof. Luc Van Gool.

Unimotion: Unifying 3D Human Motion Synthesis and Understanding

Accepted at 3DV 2025

Project Page / Paper / Code

A Continual Deepfake Detection Benchmark

Presented at WACV 2023

Project Page / Paper / Code

3D Human Body Shape and Pose Estimation

Incorporated background segmentation into 3D pose estimation to improve accuracy in complex scenes.

Object Instance Matching from Partial Data

Developed a method using DeepSDF for matching partial point clouds by learning meaningful latent shape representations.

Incremental Structure-less SfM using Trifocal Tensors

Implemented a Structure-from-Motion method using triplet optimization, improving camera pose estimation accuracy.