Question of Fig2 in main paper

[zihui0930]

Hi @Yujia-Zhang0913,

Thanks for your great work!
I'm currently studying the paper and code of Utonia. And I'm particularly interested in the conclusion of Fig.2 in the main paper.
Fig.2 indicates that the Utonia features of a CAD car are highly similar to the cars in LIDAR scene data.
I'm curious how these two sets of features are extracted? Do they have different inference configurations, as the different domains, like objects and scenes, have different inference configurations in Utonia.
And it would be greatly appreciated if you could release the code of Fig.2.

Best regards,
Zihui

[Yujia-Zhang0913]

Hi
Yes, they have different inference configs. We then compute the similarity of the output features using demo 2 script.

As for the code, you can try it yourself first. I may release the demo in the next few weeks.