Multi-scale Orderless Pooling of Deep Convolutional Activation Features. Gong, Y., Wang, L., Guo, R., & Lazebnik, S. In Fleet, D. J, Pajdla, T., Schiele, B., & Tuytelaars, T., editors, Computer Vision - ECCV 2014 - 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part VII, pages 392–407, 2014. Springer.
Multi-scale Orderless Pooling of Deep Convolutional Activation Features [link]Paper  doi  abstract   bibtex   
Deep convolutional neural networks (CNN) have shown their promise as a universal representation for recognition. However, global CNN activations lack geometric invariance, which limits their robustnes
@inproceedings{Gong:2014jk,
author = {Gong, Yunchao and Wang, Liwei and Guo, Ruiqi and Lazebnik, Svetlana},
title = {{Multi-scale Orderless Pooling of Deep Convolutional Activation Features}},
booktitle = {Computer Vision - ECCV 2014 - 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part VII},
year = {2014},
editor = {Fleet, David J and Pajdla, Tom{\'a}s and Schiele, Bernt and Tuytelaars, Tinne},
pages = {392--407},
publisher = {Springer},
annote = {Essentially, it's a Spatial Pyramid Matching method, using FC7 features....

In Section 3, they argued why FC7 has some spatial specificity. I think this is consistent with later works, such as "Understanding Deep Image Representations by Inverting Them".},
keywords = {deep learning},
doi = {10.1007/978-3-319-10584-0_26},
language = {English},
read = {Yes},
rating = {3},
date-added = {2017-05-05T20:06:51GMT},
date-modified = {2017-05-05T20:42:11GMT},
abstract = {Deep convolutional neural networks (CNN) have shown their promise as a universal representation for recognition. However, global CNN activations lack geometric invariance, which limits their robustnes},
url = {http://dx.doi.org/10.1007/978-3-319-10584-0_26},
local-url = {file://localhost/Users/yimengzh/Documents/Papers3_revised/Library.papers3/Articles/2014/Gong/ECCV%202014%20Part%20VII%202014%20Gong.pdf},
file = {{ECCV 2014 Part VII 2014 Gong.pdf:/Users/yimengzh/Documents/Papers3_revised/Library.papers3/Articles/2014/Gong/ECCV 2014 Part VII 2014 Gong.pdf:application/pdf;ECCV 2014 Part VII 2014 Gong.pdf:/Users/yimengzh/Documents/Papers3_revised/Library.papers3/Articles/2014/Gong/ECCV 2014 Part VII 2014 Gong.pdf:application/pdf}},
uri = {\url{papers3://publication/doi/10.1007/978-3-319-10584-0_26}}
}

Downloads: 0