Congratulations to Seunghui, Jaeyun, and Sundong 

[Title]
DRIM: Depth Restoration with Interference Mitigation in Multiple LiDAR Depth Cameras

[Journal]
IEEE Robotics and Automation Letters (RA-L) 2025

[Authors]
Seunghui Shin, Jaeyun Jang, Sundong Park, and Hyoseok Hwang*

[Project Link]

Link: https://sites.google.com/view/drim-dataset/ 

[Summary]
LiDAR depth cameras are widely used for accurate depth measurements in numerous studies. However, a drawback of these cameras is that interference occurs in multiple camera systems, resulting in artifacts in the depth data. These artifacts pose challenges for restoration using existing image restoration methods. In this paper, we propose a novel approach, DRIM for depth restoration. Our method begins to distinguish between artifacts of interfered depth. We then propose a method that predicts these artifacts, leveraging them to restore depth. Previously, no dataset was available for learning interference in multiple LiDAR depth cameras. Therefore, we create and provide a depth interference dataset for the first time. Our experiments demonstrate superior depth restoration performance compared to other image restoration methods, and we show the capability to restore depth in challenging scenarios. Through ablation studies, we confirm that classifying and utilizing artifacts is efficient for depth restoration. Also, we demonstrate the effectiveness of the modules comprising the architecture. These results demonstrate that our proposed method effectively restores interfered depth in multiple LiDAR depth cameras.

[Key Figure]