You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have gone through both Joint discriminative and generative learning for person re-identification and A discriminatively learned CNN embedding for person re-identification, but neither of them reports a lot of the newer trained models scores (e.g. swin transformer, HRNet-18, PCB).
As far as I understand, the models provided in here are all based on the same architecture and what changes is the backbone and some parameters. For example:
all tricks means -> --warm_epoch 5 --stride 1 --erasing_p 0.5 --batchsize 8 --lr 0.02
circle loss
DG
My questions are:
Are there any other significant changes when implementing swin transformer or other backbones?
Should I use Verif-Identif [55] as model name when citing the results of this repository, regardless of the architecture? Or is there a specific notation used when citing the models from this repo?
The text was updated successfully, but these errors were encountered:
I have gone through both Joint discriminative and generative learning for person re-identification and A discriminatively learned CNN embedding for person re-identification, but neither of them reports a lot of the newer trained models scores (e.g. swin transformer, HRNet-18, PCB).
As far as I understand, the models provided in here are all based on the same architecture and what changes is the backbone and some parameters. For example:
My questions are:
The text was updated successfully, but these errors were encountered: