Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

UnFlow: Unsupervised Learning of Optical Flow with a Bidirectional Census Loss

About

In the era of end-to-end deep learning, many advances in computer vision are driven by large amounts of labeled data. In the optical flow setting, however, obtaining dense per-pixel ground truth for real scenes is difficult and thus such data is rare. Therefore, recent end-to-end convolutional networks for optical flow rely on synthetic datasets for supervision, but the domain mismatch between training and test scenarios continues to be a challenge. Inspired by classical energy-based optical flow methods, we design an unsupervised loss based on occlusion-aware bidirectional flow estimation and the robust census transform to circumvent the need for ground truth flow. On the KITTI benchmarks, our unsupervised approach outperforms previous unsupervised deep networks by a large margin, and is even more accurate than similar supervised methods trained on synthetic datasets alone. By optionally fine-tuning on the KITTI training data, our method achieves competitive optical flow accuracy on the KITTI 2012 and 2015 benchmarks, thus in addition enabling generic pre-training of supervised networks for datasets with limited amounts of ground truth.

Simon Meister, Junhwa Hur, Stefan Roth• 2017

Related benchmarks

TaskDatasetResultRank
Optical Flow EstimationKITTI 2015 (train)
Fl-epe8.1
446
Optical Flow EstimationMPI Sintel Final (train)
Endpoint Error (EPE)7.91
209
Optical FlowMPI Sintel Clean (test)
AEE9.38
158
Optical FlowMPI-Sintel final (test)
EPE7.91
137
Optical Flow EstimationSintel Final (test)
EPE10.21
133
Optical Flow EstimationSintel clean (test)
EPE9.38
120
Optical FlowKITTI 2012 (train)--
115
Optical FlowKITTI 2015 (test)
Fl Error (All)11.11
109
Optical Flow EstimationKITTI 2015 (test)
Fl-all23.3
108
Optical FlowSintel Final (train)
EPE7.91
106
Showing 10 of 16 rows

Other info

Follow for update