Open ForawardStar opened 4 years ago
The pretrained model locates at run-*/models/final.pth
.
For each link, there is only one pretrained model, does that mean this pretrained model is trained on all 6 datasets rather than one single dataset?
No, all the models w/o joint training are trained on the DUTS-TR dataset only. The details are well described in the experimental section of the paper.
yuzheyyyy @.***> 于2021年5月24日周一 下午5:53写道:
For each link, there is only one pretrained model, does that mean this pretrained model is trained on all 6 datasets rather than one single dataset?
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/backseason/PoolNet/issues/72#issuecomment-846927427, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABYO2OFLO4XW6TCCH74CMP3TPIO3PANCNFSM4T3LY7CA .
--
From BackSeason-刘姜江 Major in Computer Science and Technology Nankai University, Tianjin,300350 P.R.China @. @.>
Hi, a very great project. However, when I click the link of pretrained models, I found there are many saliency detection results, instead of the pretrained models, no matter baiduyun or google drive. So can you releas your pretrained models? Thank you.