search

Eaphan / UPIDet

Unleash the Potential of Image Branch for Cross-modal 3D Object Detection [NeurIPS2023]
Apache License 2.0
54 stars 7 forks source link

If I don't want to perform a two-dimensional segmentation task, how should I adjust the code? #7

Closed 0neDawn closed 1 year ago

0neDawn commented 1 year ago

Dear author! Thank you very much for the code you provided. I would like to ask, I do not want to perform two-dimensional segmentation tasks (segmentation tasks are temporarily useless to me and will increase training time). How should I adjust in the code?

Eaphan commented 1 year ago

Hi. During the training, it's not necessary to compute the get_segmentation_loss int point_head_vote.py. You can also forbid the image_final_head in _3DSSD_Backbone.

0neDawn commented 1 year ago

Hi. During the training, it's not necessary to compute the get_segmentation_loss int point_head_vote.py. You can also forbid the image_final_head in _3DSSD_Backbone.

Thank you for your reply. now, I know what to do.