search

ShoufaChen / DiffusionDet

[ICCV2023 Best Paper Finalist] PyTorch implementation of DiffusionDet (https://arxiv.org/abs/2211.09788)
Other
2.1k stars 162 forks source link

About the N_train problem mentioned in the paper #76

Closed XiaoyuZxy00 closed 1 year ago

XiaoyuZxy00 commented 1 year ago

Hello author, thanks for your work I know that the N_eval setting of the inference stage is set by cfg.MODEL.DiffusionDet.NUM_PROPOSALS But I haven't found the question about the N_train setting during training. Please tell me how to change N_train? If you could answer my question, I would be very grateful!

ShoufaChen commented 1 year ago

Hi,

The cfg.MODEL.DiffusionDet.NUM_PROPOSALS at the training stage represents $N_{train}$.

XiaoyuZxy00 commented 1 year ago

Hi,

The cfg.MODEL.DiffusionDet.NUM_PROPOSALS at the training stage represents Ntrain.

Thank you for your reply! How to set the parameter of N_eval?

ShoufaChen commented 1 year ago

The cfg.MODEL.DiffusionDet.NUM_PROPOSALS represents $N{train}$ at the training stage and $N{eval}$ at the inference stage. Therefore, both $N{train}$ and $N{eval}$ are set by cfg.MODEL.DiffusionDet.NUM_PROPOSALS.

XiaoyuZxy00 commented 1 year ago

Thank you for your patient answer, I have understood! Have a good weekend!