earhian

https://drive.google.com/file/d/1hfOu3_JR0vWJkNlRhKwhqJDaF3ID2vRs/view?usp=sharing

我都没注意到这个细节，我当时参考某个行人reid里的细节

step 1.        freeze = False        model_name = 'senet154'        min_num_class = 10        checkPoint_start = 0        lr = 3e-4        #until train map5 >= 0.98 step 2.        freeze = True        model_name = 'senet154'       ...

Could you print dist_mat.shape and is_pos.shape? By the way, you can refer to https://github.com/huanghoujing/AlignedReID-Re-Production-Pytorch/blob/master/aligned_reid/model/loss.py.

mask是我队友venn做的，用u-net实现的，参考https://github.com/SeuTao/TGS-Salt-Identification-Challenge-2018-_4th_place_solution @wangvenn

啊？ 我不是早就发了吗？ 你看看你邮箱。。

@huazai-1994 是其他人标注的，发在了讨论区。

updated it

因为我对特征做了一个归一化，值都会比较小，同时也不想让参数的范围过大，所以就乘了一个系数