I am a second-year master's student in Data and Artificial Intelligence at Institut Polytechnique de Paris (IP Paris). I have been fortunate to work under the supervision of Prof. Maks Ovsjanikov at École Polytechnique, Amal Dev Parakkat at Télécom Paris and Prof. Yi Li at IIIS, Tsinghua University (THU), My research interests lie at the intersection of 3D Computer Vision and Computer Graphics.
*Equal Contribution †Corresponding Author
TL;DR: We introduce 3DHarnessBench, a benchmark for agentic 3D-to-code reconstruction that evaluates how frontier VLMs exploit visual and geometric evidence to reconstruct a target object.
TL;DR: 4DSegStreamer is a new framework that enables existing segmentation methods real-time capable, balancing efficiency and accuracy for dense perception in dynamic environments.
TL;DR: TACO is a large-scale bimanual hand-object manipulation dataset covering extensive tool-action-object combinations in real-world scenarios, supporting test-time generalization to unseen object geometries and novel behavior triplets and benchmarking various generalizable research topics.
