GyroFlow: Gyroscope-Guided Unsupervised Optical Flow Learning

TL;DR

GyroFlow introduces a novel unsupervised optical flow learning method that integrates gyroscope data to improve accuracy in challenging scenes like fog, rain, and night, outperforming existing methods.

Contribution

This work is the first to fuse gyroscope data with image content in deep learning-based optical flow estimation, enhancing performance in difficult conditions.

Findings

Outperforms state-of-the-art methods in regular scenes

Effective in challenging scenes such as fog, rain, and night

Provides a new dataset for optical flow in diverse conditions

Abstract

Existing optical flow methods are erroneous in challenging scenes, such as fog, rain, and night because the basic optical flow assumptions such as brightness and gradient constancy are broken. To address this problem, we present an unsupervised learning approach that fuses gyroscope into optical flow learning. Specifically, we first convert gyroscope readings into motion fields named gyro field. Second, we design a self-guided fusion module to fuse the background motion extracted from the gyro field with the optical flow and guide the network to focus on motion details. To the best of our knowledge, this is the first deep learning-based framework that fuses gyroscope data and image content for optical flow learning. To validate our method, we propose a new dataset that covers regular and challenging scenes. Experiments show that our method outperforms the state-of-art methods in both regular and challenging scenes. Code and dataset are available at https://github.com/megvii-research/GyroFlow.