RNN Fisher Vectors for Action Recognition and Image Annotation
Abstract
Recurrent Neural Networks (RNNs) have had considerable success in classifying and predicting sequences. We demonstrate that RNNs can be effectively used in order to encode sequences and provide effective representations. The methodology we use is based on Fisher Vectors, where the RNNs are the generative probabilistic models and the partial derivatives are computed using backpropagation. State of the art results are obtained in two central but distant tasks, which both rely on sequences: video action recognition and image annotation. We also show a surprising transfer learning result from the task of image annotation to the task of video action recognition.
Cite
Text
Lev et al. "RNN Fisher Vectors for Action Recognition and Image Annotation." European Conference on Computer Vision, 2016. doi:10.1007/978-3-319-46466-4_50Markdown
[Lev et al. "RNN Fisher Vectors for Action Recognition and Image Annotation." European Conference on Computer Vision, 2016.](https://mlanthology.org/eccv/2016/lev2016eccv-rnn/) doi:10.1007/978-3-319-46466-4_50BibTeX
@inproceedings{lev2016eccv-rnn,
title = {{RNN Fisher Vectors for Action Recognition and Image Annotation}},
author = {Lev, Guy and Sadeh, Gil and Klein, Benjamin Eliot and Wolf, Lior},
booktitle = {European Conference on Computer Vision},
year = {2016},
pages = {833-850},
doi = {10.1007/978-3-319-46466-4_50},
url = {https://mlanthology.org/eccv/2016/lev2016eccv-rnn/}
}