IEEE RA-L

How to Relieve Distribution Shifts in Semantic Segmentation for Off-Road Environments

如何缓解越野环境语义分割中的分布偏移

Ji-Hoon Hwang, Daeyoung Kim, Hyung-Suk Yoon, Dong-Wook Kim, Seung-Woo Seo

机构 * Department of Electrical and Communication Engineering, Seoul National University（电子与通信工程系，首尔国立大学） 纠错

AI总结 提出ST-Seg框架，通过风格扩展和纹理正则化缓解越野场景中源-目标域差异和传感器退化导致的分布偏移，提升语义分割鲁棒性。

Comments 8 pages, 6 figures. Accepted to IEEE Robotics and Automation Letters (RA-L)

LatentAM: Real-Time, Large-Scale Latent Gaussian Attention Mapping via Online Dictionary Learning

LatentAM：通过在线字典学习实现实时、大规模的潜在高斯注意力映射

Junwoon Lee, Yulun Tian

机构 * Robotics Department, University of Michigan（密歇根大学机器人系） 纠错

AI总结 LatentAM通过在线字典学习实现实时、大规模的潜在高斯注意力映射，提升机器人感知的特征重建保真度与实时性

Comments 8 pages, 7 figures. Accepted for RA-L. Homepage: this https URL (https://junwoonlee.github.io/projects/LatentAM/) Code: this https URL (https://github.com/UMich-SSI-Lab/latentam)