Xiaohang (Joshua) Yang

Intern Resercher
3DNR Group, Autonomous Driving Group, Shanghai AI Lab, Shanghai
Office: L1-12F-C, Shanghai AI Lab

Hi there! This is Xiaohang (Joshua) Yang. I am currently a research intern at 3D Neural Rendering Group, Shanghai AI Lab. My research interests lie in neural rendering, 3D reconstruction, video processing and perception. In particular, I study categorical neural representations and neural rendering of dynamic scenes, frame interpolation, action recognition, and optical flow estimation.

Previously, I got my Master’s and Bachelor’s degree in Control Science and Engineering (Computer Vision and Pattern Recognition Direction) at Shanghai Jiao Tong University (SJTU). During my postgraduate, I joined the Institute of Image Processing and Pattern Recognition and worked with Prof. Jie Yang.

News

Sep 30, 2022 Homepage rebuilt. :sparkles:

Education

2019 - 2022 M.S. in Control Science and Engineering, Shanghai Jiao Tong University (SJTU)
2015 - 2019 B.S. in Automation, Shanghai Jiao Tong University (SJTU)

Work Experience

Nov 2021 - Present Research Intern, 3D Neural Rendering Group, Autonomous Driving Group, Shanghai AI Lab.
Sep 2019 - Jun 2022 Research Assistant, Institute of Image Processing and Pattern Recognition, Shanghai Jiao Tong University (SJTU), supervised by Prof. Jie Yang.

Publications

  1. Graduate Thesis
    Efficient Frame Interpolation Algorithm
    Xiaohang Yang
    Master Degree Thesis (Graduate Thesis) , 2022
  2. ICASSP
    Unsupervised Motion Representation Enhanced Network for Action Recognition
    Xiaohang Yang, Lingtong Kong, and Jie Yang
    International Conference on Acoustics, Speech and Signal Processing (ICASSP) , 2021
  3. ICASSP
    Oas-net: Occlusion aware sampling network for accurate optical flow
    Lingtong Kong, Xiaohang Yang, and Jie Yang
    International Conference on Acoustics, Speech and Signal Processing (ICASSP) , 2021
  4. arXiv
    Homography Estimation Network Based on Dense Correspondence
    Xiaohang Yang, Lingtong Kong, Ziyun Liang, and Jie Yang
    arXiv (arXiv) , 2021

Projects

Categorical and Deformable Neural Representation

Pedestrian Reconstruction.
Vehicle Reconstruction.

Neural Rendering of Dynamic Scenes

Reproduce the Scene with Fixed Time.
Reproduce the Scene with Fixed Viewpoint.

Video Frame Interpolation

Original Video Frames.
8x Interpolation & Smoother Motion.

Optical Flow & Action Recognition

Optical Flow Estimation on KITTI.
Action Recognition Enhanced by Unsupervised Optical Flow.
Please refer to the Projects Page for more details.