FlowIt: Global Matching for Optical Flow with Confidence-Guided Refinement

View PDF HTML (experimental)

Abstract:We present FlowIt, a novel architecture for optical flow estimation designed to robustly handle large pixel displacements. At its core, FlowIt leverages a hierarchical transformer architecture that captures extensive global context, enabling the model to effectively model long-range correspondences. To overcome the limitations of localized matching, we formulate the flow initialization as an optimal transport problem. This formulation yields a highly robust initial flow field, alongside explicitly derived occlusion and confidence maps. These cues are then seamlessly integrated into a guided refinement stage, where the network actively propagates reliable motion estimates from high-confidence regions into ambiguous, low-confidence areas. Extensive experiments across the Sintel, KITTI, Spring, and LayeredFlow datasets validate the efficacy of our approach. FlowIt achieves state-of-the-art results on the competitive Sintel and KITTI benchmarks, while simultaneously establishing new state-of-the-art cross-dataset zero-shot generalization performance on Sintel, Spring, and LayeredFlow.

Subjects:

Computer Vision and Pattern Recognition (cs.CV)

Cite as: arXiv:2603.28759 [cs.CV]

(or arXiv:2603.28759v1 [cs.CV] for this version)

https://doi.org/10.48550/arXiv.2603.28759

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Sadra Safadoust [view email] [v1] Mon, 30 Mar 2026 17:58:12 UTC (29,467 KB)