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 LiDAR】PIXOR : BEVで3D Object Detection

Posted at

PIXOR: Real-time 3D Object Detection from Point Clouds

![image.png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/8dafa547-efe6-4579-fefe-6a76c02d1c60.png)

アルゴリズム

入力

![image.png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/6bd8af5f-e5d1-03c3-a4b2-3d7c3ecd5fdc.png) 2DのGridに区切り、(H/dH+3)のChannelを持つ

H/dHは、dHの間隔で高さ方向に区切られた領域に点が存在するか(occupancy)
3は、高さを3つに区切った、領域のintencity*
*たぶん。。。

ネットワーク

![image.png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/80566ec5-0568-3668-e934-7ccd07096e0b.png) ただたんに2DのObject DetectionをBEVで行うだけ。 LocalとGlobalの情報得るために、FPN(Feature Pyramid Network)を使うと良いらしい。

出力

![image (1).png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/43739d95-12c7-b780-83c6-db5074cdff29.png) 高さ方向の情報は使わないから、zcとhは出力しないらしい。

結果

![image.png](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/482094/e0bddc1f-62c0-71f1-ec7e-6c8b564f6ac3.png) ・2Dだけに、なかなか早くなってる

結論

・かなりシンプルだし入力を変えたり、ネットワークを深くしたりと色々工夫すればもう少し精度あがりそう。

参考文献

PIXOR: Real-time 3D Object Detection from Point Clouds https://arxiv.org/pdf/1902.06326.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?