0
1

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.

【StereoDepth】CRL : stereo matchingでもtwo stageネットワーク

Last updated at Posted at 2020-10-17

Cascade Residual Learning: A Two-stage Convolutional Neural Network for Stereo Matching

![image.png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/055f5c40-5f97-e306-23ec-19853bc3fcf8.png)

Stereo Matchingでもtwo-stage Networkの方がone-stageより精度が高い!

DipFullNet

![image.png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/92b5e100-dfab-7b8b-e0f8-0ff9b09f2b94.png) DispFullNetは[FlowNetC](https://arxiv.org/pdf/1612.01925.pdf)をHalf ResolutionではなくFull Resolution(入力と同じ解像度)でoutputするように改良したモデル。 入力はLeft ImageとRight Imageで出力がFull ResolutionのDisparity Mapとなる。

DispResNet

![image.png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/b3a4478c-19b4-6ba4-e643-7c3feac8bb6b.png) IR(右の画像)とd1を使いIL'(左の推定した画像)を作る。 IL(左の画像)とIL'を使いeL(推定のエラー)を計算する。 それらをDispReNetに入力し、d1をdownsampleしたものとr2(s)を足し合わせる事でd2(s)を推定する。

image.png

これをFull Resolutionまで計算し、最終結果d2(0)を出力する。

結果

![image.png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/3c8f4ea2-ee1b-f51a-af02-14f6b3994cfa.png) two-stageのCRLは他のネットワークより詳細を学習出来ている。

参考文献

Cascade Residual Learning: A Two-stage Convolutional Neural Network for Stereo Matching https://arxiv.org/pdf/1708.09204.pdf
0
1
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
1

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?