0
0

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.

【3D OD Stereo】PSEUDO-LIDAR++ : Depth Cost VolumeとDepth Correlation

Posted at

PSEUDO-LIDAR++:ACCURATE DEPTH FOR 3D OBJECT DETECTION IN AUTONOMOUS DRIVING

Depth Cost Volume

![image.png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/02e99c18-2398-b6d9-4179-9357662859f0.png) Disparity Cost VolumeをDepth Cost Volumeに変換する。DisparityのCost Volumeに3D CNNを掛けると同じFilterでDisparity(1/Depth)をSmoothingするので、Disparityの小さい(遠い)物体が余計にSmoothingされてしまう事が指摘されてました。depth = f * b / dispの座標変換をしてから3D CNNでを掛けると精度が上がるというのに衝撃を覚えました! ![image.png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/2a009bc8-6510-a09f-78c9-5f5c56291509.png)

Depth Correlation

![image.png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/913d82c9-6dbd-c7e1-5e1f-ea3fac2b14ce.png) LiDARのPointを使ってStereo Depthで得たDepthマップのオフセットを修正します。

結果

![image.png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/4833189d-4cc6-fffd-ec9a-ad9f6218092c.png) Depth Cost VolumeもDepth Correctionも両方とも精度を上げている!!

結論

・Cost VolumeをDisparityからDepthに変換する事でより意味のあるSmoothingが出来る。 ・Depth Correlationをする事でDepthのOffsetを4 beamのLiDARで修正する事が出来た。

参考文献

PSEUDO-LIDAR++: ACCURATE DEPTH FOR 3D OBJECT DETECTION IN AUTONOMOUS DRIVING https://arxiv.org/pdf/1906.06310.pdf
0
0
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
0

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?