0
2

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?

More than 3 years have passed since last update.

【Semantic Segmentation】DeepLab(v2) : DeepLab(v1)との違いは?

Last updated at Posted at 2020-11-27

背景

[DeepLab(v1)](https://qiita.com/minh33/items/8eb31d16a975d2a87de5)を調べたから次はDeepLab(v2)だ!と思い探していると、恐らくそれっぽい論文が見つかったのだが、タイトルがほぼ一緒で戸惑った。ちょっと文字足しだけやんけ

DeepLab(v1)

SEMANTIC IMAGE SEGMENTATION WITH DEEP CONVOLUTIONAL NETS AND FULLY CONNECTED CRFS

DeepLab(v2)

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution,and Fully Connected CRFs

version1からのupdate

image.png

Spatial Pyramid Poolingを知っている人なら簡単に分かると思うのが、ASPPではPoolingで解像度を落とさず、代わりにAtrous(Dilated) Convolutionで全体の特徴を取りに行くので特徴量Mapが小さくならずに済む。
downsampleしたりupsampleせずに情報が得られるから、個人的に結構すごいと思う。

image.png

Atrous(Dilated) Convolutionの凄さ

image.png

図からみて分かるように
downsample->convolution->upsampleするより
Atrous Convolutionする方が特徴がDenseに取れる。

上の方法だとSmoothingする為のLayerもう一枚追加する必要がありそう。

結論

ASPP Moduleが追加された事 これは大きな貢献だと思う

参考文献

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution,and Fully Connected CRFs https://arxiv.org/pdf/1606.00915.pdf
0
2
0

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
  3. You can use dark theme
What you can do with signing up
0
2

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?