Temporal Interpolation as an Unsupervised Pretraining Task for Optical Flow Estimation
Abstract
The difficulty of annotating training data is a major obstacle to using CNNs for low-level tasks in video. Synthetic data often does not generalize to real videos, while unsupervised methods require heuristic losses. Proxy tasks can overcome these issues, and start by training a network for a task for which annotation is easier or which can be trained unsupervised. The trained network is then fine-tuned for the original task using small amounts of ground truth data. Here, we investigate frame interpolation as a proxy task for optical flow. Using real movies, we train a CNN unsupervised for temporal interpolation. Such a network implicitly estimates motion, but cannot handle untextured regions. By fine-tuning on small amounts of ground truth flow, the network can learn to fill in homogeneous regions and compute full optical flow fields. Using this unsupervised pre-training, our network outperforms similar architectures that were trained supervised using synthetic optical flow.
Keywords
Cite
@article{arxiv.1809.08317,
title = {Temporal Interpolation as an Unsupervised Pretraining Task for Optical Flow Estimation},
author = {Jonas Wulff and Michael J. Black},
journal= {arXiv preprint arXiv:1809.08317},
year = {2018}
}
Comments
16 pages, 7 figures; accepted for publication at the German Conference for Pattern Recognition (GCPR) 2018