Zeran Ke

I am a final-year PhD student at the School of Computer Science, Wuhan University under the co-supervision of Prof. Nan Xue and Prof. Gui-Song Xia. I am currently a research intern at Ant Research.

My current research mainly focuses on 3D scene reconstruction, understanding, and generation with geometric structures.

Google Scholar  /  Github  /  Email

[03/2026]: 1 paper (LiP-MAP) is accepted to TPAMI 2026.

[01/2026]: 1 paper (Sat3DGen) is accepted to ICLR 2026 ( Rio de Janeiro, Brazil).

[10/2025]: 1 paper (Plana3r) is accepted to NeurIPS 2025 (San Diego, U.S.).

[04/2025]: 1 paper (ScaleLSD) is accepted to CVPR 2025 (Nashville, U.S.).

^* indicates equal contribution, ^† indicates the corresponding author

This paper studies the problem of 3D line mapping from a physical and topological perspective that a 3D line most naturally emerges as the edge of a finite 3D planar patch, and proposes a line–plane joint optimization framework that explicitly models learnable line and planar primitives.

Sat3DGen is a feed-forward satellite-to-3D framework that learns a structured, view-consistent NeRF-style scene from 2D satellite/street-view supervision, enabling mesh export and large-area mesh generation, surround-view video rendering, semantic-map-to-3D synthesis, and single-image DSM estimation.

This paper addresses metric 3D reconstruction of indoor scenes by using planar 3D primitives and exploiting their inherent geometric regularities with compact representations. Given two images captured from the same scene, PLANA3R outputs a set of 3D planar primitives and 6-DoF relative camera pose in metric scale.

This paper studies the problem of Line Segment Detection (LSD) in the manner of self-supervised learning for the characterization of line geometry in images, with the aim of learning a domain-agnostic robust LSD model that works well for any natural images.