ゆるふわめも

東京か京都にいます。

画像処理系のDeep Learningの基本的な手法

自分用のメモ。画像は門外漢なので必要最低限あたりを理解できたらと思います。定期的に更新予定、たぶん。

まとめ資料

www.slideshare.net

www.slideshare.net

サーベイ

The Deep Learning textbook by Ian Goodfellow and Yoshua Bengio and Aaron Courville

Deep Learning

An MIT Press book:Ian Goodfellow and Yoshua Bengio and Aaron Courville

Deep Learning in Neural Networks: An Overview

[1404.7828] Deep Learning in Neural Networks: An Overview

Schmidhuber, Jürgen. "Deep learning in neural networks: An overview." Neural Networks 61 (2015): 85-117.

88ページとかなり調べられていています。

画像・動画

画像分類問題

AlexNet (ImageNet Classification with Deep Convolutional Neural Networks)

papers.nips.cc

www.slideshare.net

AlexNet visualizationjeremykarnowski.wordpress.com

Krizhevsky, Alex, Ilya Sutskever, and Geoffrey E. Hinton. "Imagenet classification with deep convolutional neural networks." Advances in neural information processing systems. 2012.

GoogLeNet

f:id:misos:20161031020306p:plain
GoogLE Net全体図(こちらより引用)

[1409.4842] Going Deeper with Convolutions

Blog - GoogLeNet in Keras

Szegedy, Christian, et al. "Going deeper with convolutions." Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2015.

ブログが解説記事になっています。重みも公開されているのですぐ試せます。

Visual Geometry Group Network

https://arxiv.org/pdf/1409.1556.pdf

[1409.1556] Very Deep Convolutional Networks for Large-Scale Image Recognition

Visual Geometry Group Home Page

gist.github.com

Simonyan, Karen, and Andrew Zisserman. "Very deep convolutional networks for large-scale image recognition." arXiv preprint arXiv:1409.1556 (2014).

物体認識

R-CNN

github.com

Fast R-CNN

GitHub - rbgirshick/fast-rcnn: Fast R-CNN

Girshick, Ross. "Fast r-cnn." Proceedings of the IEEE International Conference on Computer Vision. 2015.

Faster R-CNN

www.slideshare.net

www.slideshare.net

github.com

github.com

Ren, Shaoqing, et al. "Faster R-CNN: Towards real-time object detection with region proposal networks." Advances in neural information processing systems. 2015.

R-CNNから順に理解していくべきなのかどうかはよく分からないです。 R-CNNは相当に遅いです。

YOLO: Real-Time Object Detection

pjreddie.com

Overfeat

code:start | CILVR Lab @ NYU

software:overfeat:start | CILVR Lab @ NYU

Sermanet, Pierre, et al. "Overfeat: Integrated recognition, localization and detection using convolutional networks." arXiv preprint arXiv:1312.6229 (2013).

next.sh

特徴抽出・エンベッディング

Caffe: Convolutional architecture for fast feature embedding

[1408.5093] Caffe: Convolutional Architecture for Fast Feature Embedding

Jia, Yangqing, et al. "Caffe: Convolutional architecture for fast feature embedding." Proceedings of the 22nd ACM international conference on Multimedia. ACM, 2014.