Atmospheric turbulence causes severe image degradation including spatially-varying blur, distortion, and intensity fluctuations due to random variations in the refractive index of air along the optical path. These degradations present a particularly difficult challenge for detecting and segmenting dynamic objects in long-range video sequences, as both the background and foreground are simultaneously affected.

UG²+ Track 3 (DOST) focuses on the problem of segmenting moving objects in turbulence-degraded video sequences. Participants are challenged to develop algorithms that can robustly detect and segment dynamic objects despite the compound difficulties of turbulence-induced distortion and the motion of the objects themselves.

Full competition details, dataset downloads, submission instructions, and evaluation guidelines are available on the official competition page on Codabench:

View Competition on Codabench

If you have any questions about this challenge track please feel free to email cvpr2026.ug2challenge@gmail.com