Learning scale-variant and scale-invariant features for deep image classification
详细信息 查看全文
- 作者：Nanne van Noord ; n.j.e.vannoord@tilburguniversity.edu" class="auth_mail" title="E-mail the corresponding authorAuthor Vitae ; Eric Postma e.o.postma@tilburguniversity.edu" class="auth_mail" title="E-mail the corresponding authorAuthor Vitae
- 关键词：Convolutional Neural Networks ; Multi-scale ; Artist Attribution ; Scale-variant Features
- 刊名：Pattern Recognition
- 出版年：2017
- 出版时间：January 2017
- 年：2017
- 卷：61
- 期：Complete
- 页码：583-592
- 全文大小：2471 K
文摘
- •
We propose a new approach for learning scale-variant and scale-invariant features.
- •
The multi-scale CNN is an ensemble of scale-specific CNN.
- •
Our multi-scale CNN is used to achieve state-of-the-art results for artist attribution.
- •
The learnt deep multi-scale representation encodes both fine and coarse characteristics.