Shengjie Zhu

Applied Scientist · Ring AI, Amazon

shngjz_selfie.jpg

Shengjie Zhu is an Applied Scientist at Amazon Ring AI team. His expertise lies in Spatial AI systems: designing customer-facing agentic front-ends grounded in a 3D-vision backend spanning camera calibration, localization, reconstruction, and Structure-from-Motion. He earned his Ph.D. from Michigan State University under the supervision of Professor Xiaoming Liu. His doctoral thesis focuses on recovering 3D structure and motion from image collections.

 Resume  Google Scholar  LinkedIn  GitHub

2026

  1. mba_vertical.gif
    Marginalized Bundle Adjustment: Multi-View Camera Pose from Monocular Depth Estimates
    Shengjie Zhu, Ahmed Abdelkader, Mark J. Matthews, and 2 more authors
    In 3DV, 2026
    PDF Code

2024

  1. ssfm.gif
    Revisit Self-supervised Depth Estimation with Local Structure-from-Motion
    Shengjie Zhu, and Xiaoming Liu
    In ECCV, 2024
    PDF Code
  2. 2011_09_26_drive_0005_sync.gif
    RePLAy: Remove Projective LiDAR Depthmap Artifacts via Exploiting Epipolar Geometry
    Shengjie* Zhu, Girish Chandar* Ganesan, Abhinav Kumar, and 1 more author
    In ECCV, 2024
    PDF Code

2023

  1. camerazoom.gif
    Tame a Wild Camera: In-the-Wild Monocular Camera Calibration
    Shengjie Zhu, Abhinav Kumar, Masa Hu, and 1 more author
    In NeurIPS, 2023
    PDF Code
  2. kitti.gif
    LightedDepth: Video Depth Estimation in Light of Limited Inference View Angles
    Shengjie Zhu, and Xiaoming Liu
    In CVPR, 2023
    PDF Code
  3. pmatch.png
    PMatch: Paired Masked Image Modeling for Dense Geometric Matching
    Shengjie Zhu, and Xiaoming Liu
    In CVPR, 2023
    PDF Code

2020

  1. zhu_brazil_liu_cvpr2020.gif
    The edge of depth: Explicit constraints between segmentation and depth
    Shengjie Zhu, Garrick Brazil, and Xiaoming Liu
    In CVPR, 2020
    PDF Code