Binh-Son Hua

Research Scientist, VinAI

 binhson.hua (at) gmail

 Google Scholar




I am a Research Scientist at VinAI, the first AI research lab in Vietnam with a focus on fundamental research.


My current research interests are in the intersection of computer vision and computer graphics, where I focus on developing deep learning techniques for effective 3D deep learning and realistic image synthesis.


Before 2020, I worked as a postdoc with Dr. Toshiya Hachisuka for physically based rendering at the University of Tokyo, with Dr. Sai-Kit Yeung for 3D deep learning at Singapore University of Technology and Design. I completed my PhD at National University of Singapore in 2015, under the supervision of Dr. Kok-Lim Low.


News

  • Oct 11, 2022: Two papers accepted to WACV 2023 (GAN inversion for 3D point clouds, single-image HDR reconstruction) and a paper accepted to ACCV 2022 (self-supervised point cloud learning with multiple-view rendering).

  • July 04, 2022: Four papers accepted to ECCV 2022 with the following topics: neural scene decoration, 4D point cloud reconstruction, text-to-image synthesis, and few-shot video classification.

  • June 17, 2022: A paper about style transfer for architectural photographs accepted to ICCP 2022.

  • Mar 26, 2022: Two co-author papers about Monte Carlo rendering accepted to SIGGRAPH North America 2022.

  • Mar 03, 2022: A paper on GAN inversion accepted to CVPR 2022.

  • Feb 14, 2022: An extended version of RIConv++, a simple rotation invariant convolution for point clouds, accepted to IJCV journal 2022.

  • Oct 20, 2021: A co-author paper on neural sequence transformation was presented at Pacific Graphics 2021 (Computer Graphics Forum).

  • Sep 29, 2021: A paper on few-shot learning accepted at NeurIPS 2021.

  • July 23, 2021: Two papers on 3D point cloud learning accepted at ICCV 2021.

  • June 12, 2021: Two papers on medical segmentation are accepted at MICCAI 2021.

  • Jan 13, 2021: A paper on network pruning (that retraining with proper learning rate schedules matters) is accepted at ICLR 2021!

Research

Last update: Oct 27, 2022