YOLOF icon indicating copy to clipboard operation
YOLOF copied to clipboard

Published 20 hours ago • verified publisher icon megvii-model

request for yolof-dc5 model

Open emjay73 opened this issue 4 years ago • 4 comments

Can I get a model config and pre-trained weights of the model denoted in this table? image

emjay73 avatar May 13 '21 00:05 emjay73

Hi, we do not have a plan to provide the 3-rd stage model now. Maybe next month. I'll keep this issue open.

You can train the third stage on your own with the provided 2-nd stage pre-trained model. Please follow the steps in this repo.

chensnathan avatar May 13 '21 03:05 chensnathan

You mean the only difference between the YOLOF_CSP_D_53_DC5_9x_stage2_3x and YOLOF-DC5 in the above table is whether the SWA phase is applied or not? Then where does the fps difference come from?

emjay73 avatar May 13 '21 05:05 emjay73

You can check the caption of the Table. When comparing with YOLOv4, we first fuse conv and bn, then keep cudnn_benchmark as True, and use half-precision for inference.

chensnathan avatar May 13 '21 06:05 chensnathan

Which means,

SOLVER.AMP.ENABLED : true 
CUDNN_BENCHMARK : true

in the config file?

emjay73 avatar May 31 '21 07:05 emjay73