search

jayLEE0301 / vq_bet_official

Official code for "Behavior Generation with Latent Actions" (ICML 2024 Spotlight)
https://sjlee.cc/vq-bet/
MIT License
89 stars 10 forks source link

Run NuScenes Driving Dataset #6

Closed x2ss closed 1 week ago

x2ss commented 3 weeks ago

Hi!

Thank you for sharing this excellent work.

I'm interested in using the NuScenes driving dataset, but I can't locate any guides.

Could you please give me some advice?

Thank you and best regards!

jayLEE0301 commented 2 weeks ago

Sorry for the late reply.

You could find how we modified vq-bet for NuScenes driving task in the last page of Appendix (https://arxiv.org/pdf/2403.03181) and also, we downloaded and used pre-processed data from GPT-Driver (https://github.com/PointsCoder/GPT-Driver)

Thank you!

x2ss commented 2 weeks ago

Sorry for the late reply.

You could find how we modified vq-bet for NuScenes driving task in the last page of Appendix (https://arxiv.org/pdf/2403.03181) and also, we downloaded and used pre-processed data from GPT-Driver (https://github.com/PointsCoder/GPT-Driver)

Thank you!

Thank you for your response.

I wonder if there is any code related to the NuScenes Driving Dataset on this Repository?

Thank you and best regards!

jayLEE0301 commented 1 week ago

There's no some code for NuScenes Driving Dataset, but the only modification I did on GPT-Driver pre-processed data is reordering the objects in the order of distance from the agent, and extracting current pose / future traj / adding one-hot object class.

I modified gpt-driver/prompt_message.py, gpt-driver/create_data.py, and gpt-driver/create_data_uniad.py

Thank you Seungjae.

x2ss commented 1 week ago

There's no some code for NuScenes Driving Dataset, but the only modification I did on GPT-Driver pre-processed data is reordering the objects in the order of distance from the agent, and extracting current pose / future traj / adding one-hot object class.

I modified gpt-driver/prompt_message.py, gpt-driver/create_data.py, and gpt-driver/create_data_uniad.py

Thank you Seungjae.

Thanks for your kind replay, it helps a lot.