Deep learning from temporal coherence in video

Hossein Mobahi, Ronan Collobert, Jason WestonICML 2009 · OpenReview

Abstract

This work proposes a learning method for deep architectures that takes advantage of sequential data, in particular from the temporal coherence that naturally exists in unlabeled video recordings. That is, two successive frames are likely to contain the same object or objects. This coherence is used as a supervisory signal over the unlabeled data, and is used to improve the performance on a supervised task of interest. We demonstrate the effectiveness of this method on some pose invariant object and face recognition tasks.
BibTeX
@inproceedings{mobahi2009deep,
  author = {Hossein Mobahi and Ronan Collobert and Jason Weston},
  title = {{Deep learning from temporal coherence in video}},
  booktitle = {International Conference on Machine Learning},
  year = {2009},
  url = {https://openreview.net/forum?id=Bk4GnjZOWH},
}

Loading related content...

Figure