how to visualize the results

[doraemonaaaa]

I found the output format is ['meta', 'plan_results', 'plan_gts'], unlike VAD's ouput ['meta', 'results', 'map_results', 'plan_results'], so when using visualization.py, it doesn't work, would you have script to visualize the results.

I am looking forward to your reply, Thanks.

[PeidongLi]

Our model doesn't output detection/mapping results. You can just obtain the missing keys from GT for visualization.

[doraemonaaaa]

Thanks for your response! Do you mean that there are no perception results and the focus is solely on planning? If we want to visualize perception, can we only use GT data? Also, is there a visualization.py script available for this? Thanks!

[PeidongLi]

Yes, you can just visualize the GT by official toolkit.

[doraemonaaaa]

Yes, you can just visualize the GT by official toolkit.

Thank you for your reply. I tried using tools/analysis_tool/visualization.py, but I found it works more like VAD — it requires perception results. Regarding the visualization results shown in your paper, are they based on VAD's perception results or ground truth data?
Is there any code available to visualize the results as shown in your paper?

[Gmonster-24]

@doraemonaaaa Did you find the code for the visual results presented in the paper?

[doraemonaaaa]

@doraemonaaaa Did you find the code for the visual results presented in the paper?

NO, I just vis it by code write by myself

[Gmonster-24]

Thank you for your reply. I would like to ask how this background was generated. Is it a BEV feature map？

[doraemonaaaa]

Thank you for your reply. I would like to ask how this background was generated. Is it a BEV feature map？

Yes, I generated bev heat map, and it look similar to this picture

[Gmonster-24]

If it's convenient, could you please share the code？ Or could you provide some reference projects? Thanks a lot! My email: 2963949023@qq.com

[doraemonaaaa]

If it's convenient, could you please share the code？ Or could you provide some reference projects? Thanks a lot! My email: 2963949023@qq.com

sorry, our work have not been accepted, if accepted, you may can find code in it, you can also follow our work, Learning to Drive with Two Minds: A Competitive Dual-Policy Approach in Latent World Models