search

bb12346 / GaitGL

40 stars 9 forks source link

Some problem about the details #1

Closed Jingwei-Liao closed 2 years ago

Jingwei-Liao commented 3 years ago

hello，we re-implement your code and we can't reach the performance you show in the paper (example, 86.0 in ST-NM), can you show more detail between feature mapping and loss function which we find the code and paper are different

bb12346 commented 3 years ago

hi, this code is corresponding to [A], which is used to HID2020 [B].

[A] Learning Effective Representations from Global and Local Features for Cross-View Gait Recognition [B] http://hid2020.iapr-tc4.org/hid-2020-competition-results/

Jingwei-Liao commented 3 years ago

We find the difference between the HID2020 and your ICCV21 paper is the loss function, but your ICCV21 paper doesn't show the detail about the loss function, can you show me more? Thanks very much !

bb12346 commented 3 years ago

We will release our code (ICCV version) as soon as possible. :)

Jingwei-Liao commented 3 years ago

oh, thanks, will the iccv code still release on this account?

bb12346 commented 3 years ago

I think I can send you the details of the loss function by email. Can you provide me with your email？

bb12346 commented 3 years ago

oh, thanks, will the iccv code still release on this account?

Yes.

Jingwei-Liao commented 3 years ago

Thinks for your help! This is my email 3222413873@qq.com