I’m a PhD student at TU Delft working on artificial intelligence for human motion understanding and social interaction. My research explores how models can extract, represent, and reason over human behavior using multimodal signals such as video and 2D/3D pose.
I design generative and transformer-based models that capture fine-grained, temporally coherent human motion, with a particular focus on multi-person settings. By leveraging structured representations of the body, I aim to improve how machines interpret actions, anticipate future behavior, and understand interactions between people.More broadly, I am interested in building AI systems that can reason about humans in context and operate effectively in socially complex environments.
Previsouly I have worked as a computer vision research intern at Bosch and BMW, obtained my MSc cum laude & obtained my BSc cum laude and with honours.
Human-like multimodal reaction generation is essential for natural group interactions between humans and intelligent embodied AI. However, existing approaches are often limited to single-modality or speaking-only …
Zhi-Yi Lin, Thomas Markhorst, Jouh Yeong Chew, Xucong Zhang · CVPR 2026
Multi-person social interactions are inherently built on coherence and relationships among all individuals within the group, making multi-person localization and body pose estimation essential to understanding these …
Thomas Markhorst, Zhi-Yi Lin, Joug Yeong Chew, Jan van Gemert, Xucong Zhang · CVPRw 2026
In this paper, we jointly combine image classification and image denoising, aiming to enhance human perception of noisy images captured by edge devices, like low-light security cameras. In such settings, it is important …
Thomas Markhorst, Jan van Gemert, Osman Kayhan · ECCVw 2024
