search

Haiyang-W / DSVT

[CVPR2023] Official Implementation of "DSVT: Dynamic Sparse Voxel Transformer with Rotated Sets"
https://arxiv.org/abs/2301.06051
Apache License 2.0
361 stars 28 forks source link

Question about the CT3D-based two-stage DSVT on WOD training yaml file #24

Closed ylwhxht closed 1 year ago

ylwhxht commented 1 year ago

Thanks for your excellent work, I wanted to reproduce the experimental results of the two-stage DSVT-TS on waymo's 100% training set, but I did not find the relevant two-stage configuration file with CT3D as the model, all of which are the first-stage configuration files with CenterPoint as the model.

Can you upload or tell me where it is?

Thank you very much

chenshi3 commented 1 year ago

Sorry for the late reply. Regarding the training of our two-stage model, we fix the 1st-stage DSVT model and only finetune the 2nd-stage refinement module for 12 epochs with the same training schedule. The process is a bit complex and detailed, we may not release it. However, if you require it urgently, please feel free to send me an email.

Haiyang-W commented 1 year ago

Thanks for your excellent work, I wanted to reproduce the experimental results of the two-stage DSVT-TS on waymo's 100% training set, but I did not find the relevant two-stage configuration file with CT3D as the model, all of which are the first-stage configuration files with CenterPoint as the model.

Can you upload or tell me where it is?

Thank you very much

Please give me your email. I will give you an unrefined CT3D-based two-stage code in a few days as soon as possible. :)

Haiyang

zizhengu commented 1 year ago

Hi! Thank you very much for your previous help and patience. I have achieved excellent results in my task with your amazing DSVT model. I would like to try the two-stage model further. Could you please send me the CT3D-based code ? Thanks again! Also, I would like to know if the two-stage model would significantly slow down the inference speed. Here is my email: zizhengu@163.com

Haiyang-W commented 1 year ago

Hi! Thank you very much for your previous help and patience. I have achieved excellent results in my task with your amazing DSVT model. I would like to try the two-stage model further. Could you please send me the CT3D-based code ? Thanks again! Also, I would like to know if the two-stage model would significantly slow down the inference speed. Here is my email: zizhengu@163.com

Have sent it to email! :)

Zhaoguanhua commented 7 months ago

hello, Could you send me the CT3D-based two-stage code? my email: a18707199526@163.com