U²Flow jointly estimates optical flow and uncertainty in an unsupervised recurrent setup by deriving uncertainty from augmentation consistency via Laplace maximum likelihood, then using it to refine flow and modulate smoothness.
Learning by analogy: Reliable supervision from transformations for unsupervised optical flow estima- tion
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
U$^{2}$Flow: Uncertainty-Aware Unsupervised Optical Flow Estimation
U²Flow jointly estimates optical flow and uncertainty in an unsupervised recurrent setup by deriving uncertainty from augmentation consistency via Laplace maximum likelihood, then using it to refine flow and modulate smoothness.