@inproceedings{024b62879d1f4eeea28f0afe9508bc07,
title = "Learning image matching by simply watching video",
abstract = "This work presents an unsupervised learning based approach to the ubiquitous computer vision problem of image matching. We start from the insight that the problem of frame interpolation implicitly solves for inter-frame correspondences. This permits the application of analysisby-synthesis: we first train and apply a Convolutional Neural Network for frame interpolation, then obtain correspondences by inverting the learned CNN. The key benefit behind this strategy is that the CNN for frame interpolation can be trained in an unsupervised manner by exploiting the temporal coherence that is naturally contained in real-world video sequences. The present model therefore learns image matching by simply “watching videos”. Besides a promise to be more generally applicable, the presented approach achieves surprising performance comparable to traditional empirically designed methods.",
keywords = "Analysis by synthesis, Convolutional neural network, Image matching, Temporal coherence, Unsupervised learning",
author = "Gucan Long and Laurent Kneip and Alvarez, {Jose M.} and Hongdong Li and Xiaohu Zhang and Qifeng Yu",
note = "Publisher Copyright: {\textcopyright} Springer International Publishing AG 2016.; 14th European Conference on Computer Vision, ECCV 2016 ; Conference date: 08-10-2016 Through 16-10-2016",
year = "2016",
doi = "10.1007/978-3-319-46466-4_26",
language = "English",
isbn = "9783319464657",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "434--450",
editor = "Bastian Leibe and Jiri Matas and Nicu Sebe and Max Welling",
booktitle = "Computer Vision - 14th European Conference, ECCV 2016, Proceedings",
address = "Germany",
}