Liangchen Song
Hi there! I am a final-year Ph.D. student from University at Buffalo, advised by Prof. Junsong Yuan.
Email:   lsong8 at buffalo.edu
Links:   Github / dblp / Google scholar

Projects

Photometric stereo for dynamic scenes
  • Assuming there is only one light for each frame.
  • Aiming to recover the normal map for all frames.
Neural 4D Light Field Rendering

Selected (three/five) Publications

(* indicates equal contribution)

NeRFPlayer: A Streamable Dynamic Scene Representation with Decomposed Neural Radiance Fields.

Liangchen Song, Anpei Chen, Zhong Li, Zhang Chen, Lele Chen, Junsong Yuan, Yi Xu, and Andreas Geiger.

Preprint

[pdf] [project page]

PREF: Predictability Regularized Neural Motion Fields.

Liangchen Song, Xuan Gong, Benjamin Planche, Meng Zheng, David Doermann, Junsong Yuan, Terrence Chen, and Ziyan Wu.

ECCV 2022 (oral)

[pdf] [project page]

# Implicitly representing the patterns of motion.
Unsupervised domain adaptive re-identification: Theory and practice.

Liangchen Song*, Cheng Wang*, Lefei Zhang, Bo Du, Qian Zhang, Chang Huang, and Xinggang Wang.

Pattern Recognition 2020 (3rd most cited PR 2020 paper)

[pdf] [code]

# My first work after learning deep learning. Got rejected by almost all 'top' AI conferences.
# Happy to see that some people cited this paper.
NeRFPlayer: A Streamable Dynamic Scene Representation with Decomposed Neural Radiance Fields.

Liangchen Song, Anpei Chen, Zhong Li, Zhang Chen, Lele Chen, Junsong Yuan, Yi Xu, and Andreas Geiger.

Preprint

[pdf] [project page]

PREF: Predictability Regularized Neural Motion Fields.

Liangchen Song, Xuan Gong, Benjamin Planche, Meng Zheng, David Doermann, Junsong Yuan, Terrence Chen, and Ziyan Wu.

ECCV 2022 (oral)

[pdf] [project page]

# Implicitly representing the patterns of motion.
Stacked Homography Transformations for Multi-View Pedestrian Detection.

Liangchen Song, Jialian Wu, Ming Yang, Qian Zhang, Yuan Li, and Junsong Yuan.

ICCV 2021 (oral)

[pdf] [code]

# Sometimes I cannot believe this is an oral paper.
# But I do believe this is not the worst oral paper. :)

Unsupervised domain adaptive re-identification: Theory and practice.

Liangchen Song*, Cheng Wang*, Lefei Zhang, Bo Du, Qian Zhang, Chang Huang, and Xinggang Wang.

Pattern Recognition 2020 (3rd most cited PR 2020 paper)

[pdf] [code]

# My first work after learning deep learning. Got rejected by almost all 'top' AI conferences.
# Happy to see that some people cited this paper.
Nonlocal patch based t-SVD for image inpainting: algorithm and error analysis.

Liangchen Song, Bo Du, Lefei Zhang, Liangpei Zhang, Jia Wu, and Xuelong Li.

AAAI 2018 (oral)

[pdf] [code]

# No networks involved. First objective function, then optimizing it.
# Not end-to-end. No backpropagation. Looks weird now?