Lightnews — Scholar-powered news

Zhenjun Zhao

@ericzzj.bsky.social

1.3K followers 470 following 1K posts

ericzzj1989.github.io PhD from CUHK. 3D vision, SLAM, SfM, Image Matching (https://github.com/ericzzj1989/Awesome-Image-Matching).

github.com

Posts Media Videos Starter Packs

Pinned

Zhenjun Zhao @ericzzj.bsky.social · May 9

🎉 Thrilled to share our CVPR 2025 Award Candidate & Oral paper:

🔹 GlobustVP
Convex Relaxation for Robust Vanishing Point Estimation in Manhattan World

🧱 Global optimality
💥 Tolerates up to 70% outliers
⚡ Fast runtime

📄 Paper: arxiv.org/abs/2505.04788

💻 Code: github.com/WU-CVGL/GlobustVP

1/

1 5 29

Zhenjun Zhao @ericzzj.bsky.social · 22h

G4Splat: Geometry-Guided Gaussian Splatting with Generative Prior

Junfeng Ni, @yixinchen.bsky.social, Zhifei Yang, Yu Liu, Ruijie Lu, Song-Chun Zhu, Siyuan Huang

tl;dr: planar surfaces->plane-aware depth; geometry guidance->generative refinement

arxiv.org/abs/2510.12099

Zhenjun Zhao @ericzzj.bsky.social · 22h

tl;dr: SCR training->maximum likelihood learning; depth (distribution) prior & 3D point cloud diffusion prior->SCR

Zhenjun Zhao @ericzzj.bsky.social · 22h

Scene Coordinate Reconstruction Priors

@wenjingbian.bsky.social, @axelbarroso.bsky.social, Tommaso Cavallari, Victor Adrian Prisacariu, @ericbrachmann.bsky.social

arxiv.org/abs/2510.12387

1 4

Zhenjun Zhao @ericzzj.bsky.social · 22h

E-MoFlow: Learning Egomotion and Optical Flow from Event Data via Implicit Regularization

Wenpu Li, Bangyan Liao, Yi Zhou, Qi Xu, Pian Wan, Peidong Liu

tl;dr: implicit spatial-temporal and geometric regularization->egomotion and optical flow

arxiv.org/abs/2510.12753

Zhenjun Zhao @ericzzj.bsky.social · 22h

UniGS: Unified Geometry-Aware Gaussian Splatting for Multimodal Rendering

Yusen Xie, Zhenmin Huang, Jianhao Jiao, Dimitrios Kanoulas, Jun Ma

tl;dr: differentiable ray-ellipsoid intersection; analytical gradients; learnable attribute->prune Gaussians

arxiv.org/abs/2510.12174

Zhenjun Zhao @ericzzj.bsky.social · 22h

tl;dr: align intermediate visual embeddings of VLAs with geometric representations produced by pretrained 3D foundation models

Zhenjun Zhao @ericzzj.bsky.social · 22h

Spatial Forcing: Implicit Spatial Representation Alignment for Vision-language-action Model

Fuhao Li, Wenxuan Song, Han Zhao, Jingbo Wang, Pengxiang Ding, Donglin Wang, Long Zeng, Haoang Li

arxiv.org/abs/2510.12276

1 1

Zhenjun Zhao @ericzzj.bsky.social · 22h

Uncertainty Matters in Dynamic Gaussian Splatting for Monocular 4D Reconstruction

Fengzhi Guo, Chih-Chuan Hsu, Sihao Ding, Cheng Zhang

tl;dr: time-varying per-Gaussian uncertainty->spatio-temporal graph->4DGS

arxiv.org/abs/2510.12768

Zhenjun Zhao @ericzzj.bsky.social · 1d

VA-GS: Enhancing the Geometric Representation of Gaussian Splatting via View Alignment

Qing Li, Huifang Feng, Xun Gong, Yu-Shen Liu

tl;dr: edge-aware image cues+visibility-aware photometric alignment loss+normal-based constraints+deep image feature embeddings

arxiv.org/abs/2510.11473

Zhenjun Zhao @ericzzj.bsky.social · 1d

tl;dr: transformer-based scene-agnostic coordinate regressor+map codes; separate pre-training

Zhenjun Zhao @ericzzj.bsky.social · 1d

ACE-G: Improving Generalization of Scene Coordinate Regression Through Query Pre-Training

@roym899.bsky.social, @axelbarroso.bsky.social, Tommaso Cavallari, @amonszpart.bsky.social, @scriptide.bsky.social, Victor Adrian Prisacariu, @ericbrachmann.bsky.social

arxiv.org/abs/2510.11605

1 2

Zhenjun Zhao @ericzzj.bsky.social · 1d

VG-Mapping: Variation-Aware 3D Gaussians for Online Semi-static Scene Mapping

Yicheng He, Jingwen Yu, Guangcheng Chen, Hong Zhang

tl;dr: 3DGS+TSDF-based voxel map->variation detection->variation-aware density control->decouple density updates

arxiv.org/abs/2510.09962

Zhenjun Zhao @ericzzj.bsky.social · 1d

sqrtVINS: Robust and Ultrafast Square-Root Filter-based 3D Motion Tracking

Yuxiang Peng, Chuchu Chen, Kejian Wu, Guoquan Huang

tl;dr: SR-VINS+Cholesky decomposition (LLT)-based square-root filter update+dynamic initialization

arxiv.org/abs/2510.10346

Zhenjun Zhao @ericzzj.bsky.social · 1d

Shuanghao Bai, Wenxuan Song, Jiayi Chen, Yuheng Ji, Zhide Zhong, Jin Yang, Han Zhao, Wanqi Zhou, Wei Zhao, Zhe Li, Pengxiang Ding, Cheng Chi, Haoang Li, Chang Xu, Xiaolong Zheng, Donglin Wang, Shanghang Zhang, Badong Chen

Zhenjun Zhao @ericzzj.bsky.social · 1d

Towards a Unified Understanding of Robot Manipulation: A Comprehensive Survey

tl;dr: in title

arxiv.org/abs/2510.10903

Zhenjun Zhao @ericzzj.bsky.social · 2d

MCMC: Bridging Rendering, Optimization and Generative AI

Gurprit Singh, @wjakob.bsky.social

tl;dr: in title

arxiv.org/abs/2510.09078

Zhenjun Zhao @ericzzj.bsky.social · 5d

Learning Neural Exposure Fields for View Synthesis

@miniemeyer.bsky.social, Fabian Manhardt, Marie-Julie Rakotosaona, Michael Oechsle, Christina Tsalicoglou, Keisuke Tateno, @jonbarron.bsky.social, Federico Tombari

tl;dr: neural field->optimal exposure value per 3D point

arxiv.org/abs/2510.08279

Zhenjun Zhao @ericzzj.bsky.social · 5d

tl;dr: MASt3R->pose estimation; π3->loop detection; hierarchical semi-implicit Gaussian+LoD-aware densification->mapping

Zhenjun Zhao @ericzzj.bsky.social · 5d

ARTDECO: Towards Efficient and High-Fidelity On-the-Fly 3D Reconstruction with Structured Scene Representation

Guanghao Li, Kerui Ren, Linning Xu, Zhewen Zheng, Changjian Jiang, Xin Gao, Bo Dai, Jian Pu, Mulin Yu, Jiangmiao Pang

arxiv.org/abs/2510.08551

1 1

Zhenjun Zhao @ericzzj.bsky.social · 5d

D2GS: Depth-and-Density Guided Gaussian Splatting for Stable and Accurate Sparse-View Reconstruction

Meixi Song, Xin Lin, Dizhe Zhang, Haodong Li, Xiangtai Li, Bo Du, Lu Qi

tl;dr: local density and camera distance->dropout score->overfitting; depth priors->underfitting

arxiv.org/abs/2510.08566

Zhenjun Zhao @ericzzj.bsky.social · 5d

ReSplat: Learning Recurrent Gaussian Splats

@haofeixu.bsky.social, @danielbarath.bsky.social, @andreasgeiger.bsky.social, @marcpollefeys.bsky.social

tl;dr: rendering error->recurrent network->Gaussian updates

arxiv.org/abs/2510.08575

1 3

Zhenjun Zhao @ericzzj.bsky.social · 5d

Splat the Net: Radiance Fields with Splattable Neural Primitives

Xilong Zhou, Bao-Huy Nguyen, Loïc Magne, Vladislav Golyanik, Thomas Leimkühler, Christian Theobalt

tl;dr: primitive->bounded by an ellipsoid; density->shallow neural network

arxiv.org/abs/2510.08491

Zhenjun Zhao @ericzzj.bsky.social · 6d

tl;dr: point maps from MoGe+dense correspondences from DKM->initial alignment->graph-based refinement with 3D points and normals

Zhenjun Zhao @ericzzj.bsky.social · 6d

MoRe: Monocular Geometry Refinement via Graph Optimization for Cross-View Consistency

Dongki Jung, Jaehoon Choi, Yonghan Lee, Sungmin Eum, Heesung Kwon, Dinesh Manocha

arxiv.org/abs/2510.07119

1 1

Zhenjun Zhao @ericzzj.bsky.social · 6d

UniFField: A Generalizable Unified Neural Feature Field for Visual, Semantic, and Spatial Uncertainties in Any Scene

Christian Maurer, @snehaljauhri.bsky.social, Sophie Lueth, @georgiachal.bsky.social

tl;dr: in title

arxiv.org/abs/2510.06754