Questions about result verification

[TJXA]

Great work! How do you use the generated images/videos for detection in perception models? In addition to image data, a lot of other data is usually required, such as can_bus data. How do you obtain this data?

[flymin]

Hi, for controllable generation, in general, you can just provide the model with annotation you want and take the generated image/video as sensor data. Through this process, there is no need to annotate the generation.

[TJXA]

Thank you very much for your prompt reply. I read this paper, which mentioned using generated data to train/test perception models, such as BEVFformer. From what I understand, it is not enough to just input image data, and data such as can_bus is also required. I would like to know how you obtained this data.

[flymin]

can_bus goes to the annotations in me previous reply.

[github-actions[bot]]

This issue is stale because it has been open for 7 days with no activity. If you do not have any follow-ups, the issue will be closed soon.