首次实现无监督光流与像素级不确定性联合估计，提升模型鲁棒性。
U$^{2}$Flow: Uncertainty-Aware Unsupervised Optical Flow Estimation
- 通过增广一致性构建拉普拉斯似然损失，无需真值即可学习不确定性。
- 在KITTI和Sintel上达到无监督方法最优性能，不确定性图可靠。
- 适合需要可信预测结果的自动驾驶、机器人视觉场景使用。
无监督光流方法通常缺乏可靠的不确定性估计，限制了其鲁棒性和可解释性。我们提出U$^{2}$Flow，首个递归式无监督框架，可联合估计光流与每像素不确定性。核心创新是解耦学习策略，通过拉普拉斯最大似然目标从增广一致性中获取不确定性监督，实现无需真值的稳定训练。预测的不确定性被融入网络，用于引导自适应光流优化并动态调节区域平滑损失。此外，引入不确定性引导的双向光流融合机制，增强复杂区域的鲁棒性。在KITTI和Sintel上的大量实验表明，U$^{2}$Flow在无监督方法中达到最先进性能，同时生成高度可靠的不确定性图，验证了联合估计范式的有效性。代码已公开于https://github.com/sunzunyi/U2FLOW。
原文摘要 · Abstract (English)
Unsupervised optical flow methods typically lack reliable uncertainty estimation, limiting their robustness and interpretability. We propose U$^{2}$Flow, the first recurrent unsupervised framework that jointly estimates optical flow and per-pixel uncertainty. The core innovation is a decoupled learning strategy that derives uncertainty supervision from augmentation consistency via a Laplace-based maximum likelihood objective, enabling stable training without ground truth. The predicted uncertainty is further integrated into the network to guide adaptive flow refinement and dynamically modulate the regional smoothness loss. Furthermore, we introduce an uncertainty-guided bidirectional flow fusion mechanism that enhances robustness in challenging regions. Extensive experiments on KITTI and Sintel demonstrate that U$^{2}$Flow achieves state-of-the-art performance among unsupervised methods while producing highly reliable uncertainty maps, validating the effectiveness of our joint estimation paradigm. The code is available at https://github.com/sunzunyi/U2FLOW.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。