Turbulence-Robust Dynamic Object Segmentation with Multi-Signal Priors and SAM2 Refinement 事件

Turbulence-Robust Dynamic Object Segmentation with Multi-Signal Priors and SAM2 Refinement arXiv:2605.29292v1 Announce Type: new Abstract: This technical report presents our solution for the CVPR 2026 UG2+ Challenge Track 3: Dynamic Object Segmentation in Turbulence (DOST). We design a training-free multi-signal segmentation pipeline that combines pretrained motion estimation, self-supervised semantic priors, background anomaly modeling, manually calibrated proposal fusion, and SAM2-based mask