Yiren Lu (陆弈人)

I am a PhD student in Computer Science at at Case Western Reserve University (CWRU), where I am advised by Prof. Yu Yin.

Prior to that, I received my M.S. in Computer Science and Engineering from University at Buffalo (UB) in 2024, where I was fortunate to be a member of SAIR lab and advised by Prof. Chen Wang. I received my B.Eng. in Computer Science from ShanghaiTech University supervised by Prof. Sören Schwertfeger.

I have a broad research interest in 3D Computer Vision, including 3D & 4D scene reconstruction, understanding and generation. Previously, I also worked on simultaneous localization and mapping (SLAM) and Robot Perception.

Email  /  Google Scholar  /  CV  /  LinkedIn  /  GitHub

News!

Working Experience

Selected Publications (* denotes equal contribution, full publication list)

Segment then Splat: Unified 3D Open-Vocabulary Segmentation via Gaussian Splatting

We propose Segment then Splat, a method that reverses the long established approach of “segmentation after reconstruction” by dividing Gaussians into distinct object sets before reconstruction. This approach not only eliminates semantic and geometrical ambiguity but also accelerates the optimization.

Fix False Transparency by Noise Guided Splatting

3DGS can induce “false transparency” artifacts due to its α-blending strategy. To tackle this issue, we propose Noise Guided Splatting. By injecting opaque noise Gaussians in the object volume during training, the object surfaces are encourages surface Gaussians to adopt higher opacity.

BARD-GS: Blur-Aware Reconstruction of Dynamic Scenes via Gaussian Splatting

we introduce BARD-GS, a novel approach for robust dynamic scene reconstruction that effectively handles blurry inputs and imprecise camera poses, by explicitly decomposing motion blur into camera motion blur and object motion blur to handle them separately.

View-consistent Object Removal in Radiance Fields

we introduce a novel RF editing pipeline that significantly enhances consistency by requiring the inpainting of only a single reference image. This image is then projected across multiple views using a depth-based approach, effectively reducing the inconsistencies observed with per-frame inpainting.

iSLAM: Imperative SLAM

we proposed a novel self-supervised imperative learning framework, named imperative SLAM (iSLAM), which fosters reciprocal correction between the front-end and back-end, thus enhancing performance without necessitating any external supervision.

Academic Service

Conference Reviewer

Journal Reviewer

Teaching

Teaching Assistant

Spring 2025 - CSDS 570 Deep Generative Models

• Instructor: Yu Yin

Fall 2024 - CSDS 465 Computer Vision

• Instructor: Yu Yin

Spring 2024 - CSDS 465 Computer Vision

• Instructor: Yu Yin

Design and source code from Jon Barron's website