DA-Flow: Degradation-Aware Optical Flow Estimation with Diffusion Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Min, Jaewon, Lee, Jaeeun, Choi, Yeji, Cho, Paul Hyunbin, Kim, Jin Hyeon, Lee, Tae-Young, Ahn, Jongsik, Lee, Hwayeong, Park, Seonghyun, Kim, Seungryong
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910069720875008
author Min, Jaewon
Lee, Jaeeun
Choi, Yeji
Cho, Paul Hyunbin
Kim, Jin Hyeon
Lee, Tae-Young
Ahn, Jongsik
Lee, Hwayeong
Park, Seonghyun
Kim, Seungryong
author_facet Min, Jaewon
Lee, Jaeeun
Choi, Yeji
Cho, Paul Hyunbin
Kim, Jin Hyeon
Lee, Tae-Young
Ahn, Jongsik
Lee, Hwayeong
Park, Seonghyun
Kim, Seungryong
contents Optical flow models trained on high-quality data often degrade severely when confronted with real-world corruptions such as blur, noise, and compression artifacts. To overcome this limitation, we formulate Degradation-Aware Optical Flow, a new task targeting accurate dense correspondence estimation from real-world corrupted videos. Our key insight is that the intermediate representations of image restoration diffusion models are inherently corruption-aware but lack temporal awareness. To address this limitation, we lift the model to attend across adjacent frames via full spatio-temporal attention, and empirically demonstrate that the resulting features exhibit zero-shot correspondence capabilities. Based on this finding, we present DA-Flow, a hybrid architecture that fuses these diffusion features with convolutional features within an iterative refinement framework. DA-Flow substantially outperforms existing optical flow methods under severe degradation across multiple benchmarks.
format Preprint
id arxiv_https___arxiv_org_abs_2603_23499
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle DA-Flow: Degradation-Aware Optical Flow Estimation with Diffusion Models
Min, Jaewon
Lee, Jaeeun
Choi, Yeji
Cho, Paul Hyunbin
Kim, Jin Hyeon
Lee, Tae-Young
Ahn, Jongsik
Lee, Hwayeong
Park, Seonghyun
Kim, Seungryong
Computer Vision and Pattern Recognition
Optical flow models trained on high-quality data often degrade severely when confronted with real-world corruptions such as blur, noise, and compression artifacts. To overcome this limitation, we formulate Degradation-Aware Optical Flow, a new task targeting accurate dense correspondence estimation from real-world corrupted videos. Our key insight is that the intermediate representations of image restoration diffusion models are inherently corruption-aware but lack temporal awareness. To address this limitation, we lift the model to attend across adjacent frames via full spatio-temporal attention, and empirically demonstrate that the resulting features exhibit zero-shot correspondence capabilities. Based on this finding, we present DA-Flow, a hybrid architecture that fuses these diffusion features with convolutional features within an iterative refinement framework. DA-Flow substantially outperforms existing optical flow methods under severe degradation across multiple benchmarks.
title DA-Flow: Degradation-Aware Optical Flow Estimation with Diffusion Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.23499