Semantic Scholar Open Access 2015 4561 sitasi

FlowNet: Learning Optical Flow with Convolutional Networks

Alexey Dosovitskiy P. Fischer Eddy Ilg Philip Häusser C. Hazirbas +4 lainnya

Lihat Sumber DOI

Abstrak

Convolutional neural networks (CNNs) have recently been very successful in a variety of computer vision tasks, especially on those linked to recognition. Optical flow estimation has not been among the tasks CNNs succeeded at. In this paper we construct CNNs which are capable of solving the optical flow estimation problem as a supervised learning task. We propose and compare two architectures: a generic architecture and another one including a layer that correlates feature vectors at different image locations. Since existing ground truth data sets are not sufficiently large to train a CNN, we generate a large synthetic Flying Chairs dataset. We show that networks trained on this unrealistic data still generalize very well to existing datasets such as Sintel and KITTI, achieving competitive accuracy at frame rates of 5 to 10 fps.

Topik & Kata Kunci

Computer Science

Penulis (9)

Alexey Dosovitskiy

P. Fischer

Eddy Ilg

Philip Häusser

C. Hazirbas

Vladimir Golkov

Patrick van der Smagt

D. Cremers

T. Brox

Format Sitasi

APA MLA BibTeX

Dosovitskiy, A., Fischer, P., Ilg, E., Häusser, P., Hazirbas, C., Golkov, V. et al. (2015). FlowNet: Learning Optical Flow with Convolutional Networks. https://doi.org/10.1109/ICCV.2015.316

Akses Cepat

Lihat di Sumber doi.org/10.1109/ICCV.2015.316

Informasi Jurnal

Tahun Terbit: 2015
Bahasa: en
Total Sitasi: 4561×
Sumber Database: Semantic Scholar
DOI: 10.1109/ICCV.2015.316
Akses: Open Access ✓