zkcys001/UDAStrongBaseline
Open-source stronger baseline for unsupervised or domain adaptive object re-ID.
Python
Issues
- 0
请问是否可以公布weights文件,想基于本工作在模型推理加速方面进行探索
#33 opened by mookixiao - 3
训练到一定epoch停滞的问题
#15 opened by LeeYushan - 3
请问可以使用单卡训练吗?大概需要多少显存?
#11 opened by LeeYushan - 2
训练到一定epoch之后(50+)模型输出、loss会变成nan
#25 opened by yanty123 - 3
msmt17的训练脚本没有嘛?
#27 opened by xmy0916 - 0
A question about the case of Market2MSMT
#32 opened by LeeGeon30 - 0
跑GLT II的时候出现以下问题
#31 opened by shylie728 - 0
跑GLT II 出现以下问题
#30 opened by shylie728 - 0
关于几个loss的困惑
#29 opened by Alex20172023 - 0
- 3
关于resnet50网络架构里面的_build_nonlocal函数
#24 opened by henrysun1996 - 0
- 1
请问baseline是否加了对比学习损失呢?
#19 opened by yanty123 - 9
为什么这儿的精度复现出来只有83.3%的top1,
#23 opened by henrysun1996 - 5
训练精度不对齐的问题
#20 opened by Youngboy12 - 2
关于fine-tune阶段source data 的ID loss 问题
#18 opened by yanty123 - 1
- 0
GLT_kmeans_duke2market复现很多log
#21 opened by henrysun1996 - 0
图片如此
#22 opened by henrysun1996 - 2
模型输出后,为什么要删掉一半的特征?
#17 opened by woocoder - 3
- 1
数据集问题
#12 opened by henrysun1996 - 5
- 4
Strong baseline
#14 opened by clarkhedi - 2
- 1
missing sbs_traindbscan_unc.py
#6 opened by yjh576 - 2
The result of GLT much lower than paper's
#8 opened by woocoder - 1
- 2
- 5
source domain pre-training
#5 opened by hahapt - 1
MMT results
#3 opened by SikaStar - 2
Very bad result for cross-domain reid
#2 opened by shuxjweb - 0