alibaba/cluster-contrast-reid

Issues

• batchsize大小设置问题

  #35 opened 8 months ago by BigKing992
  0

• 关于不能复现msmt17相关指标的问题

  #34 opened a year ago by hongyan-an
  1

• Why is the performance of the model trained with multi GPUs much better than that of the model trained with single GPU in this method?

  #16 opened 3 years ago by HongweiZhang97
  5

• Can't reproduce the result of MTMC17 to 33.3 using avg pooling

  #28 opened 3 years ago by haichuntai
  2

• cm模块中的backward代码好像没有起作用

  #32 opened a year ago by rrjia
  2

• 在使用nocam sampler 的情况下，batchsize越大结果反而越低

  #33 opened a year ago by zsszyx
  0

• 为啥msmt17的指标sota都只有20多呢？

  #26 opened a year ago by xmy0916
  1

• PersonX 数据集上mAP只有36.0左右

  #20 opened 3 years ago by jinx-USTC
  6

• 运行问题

  #31 opened 2 years ago by stu977
  1

• Why the result cannot be reproduced?

  #23 opened 3 years ago by tengxiao14
  2

• Why this paper is not accepted...

  #30 opened 2 years ago by dyhBUPT
  1

• dataset

  #29 opened 2 years ago by milktean
  4

• About the setting of batch_size and iter number

  #25 opened 3 years ago by xiguafeichong
  1

• Warning: Leaking Caffe2 thread-pool after fork.

  #27 opened 3 years ago by songxia928
  0

• 关于Memory Initialization的一点问题

  #15 opened 3 years ago by xmy0916
  4

• Different methods in memory update

  #24 opened 3 years ago by SuperbTUM
  0

• 实验多次复现结果不同的问题

  #7 opened 4 years ago by hu-zheng
  2

• How to determine the EPS parameters of DBSCAN for different datasets?

  #13 opened 3 years ago by lucy3589
  1

• Sorry that we didn't not provides the CM_hard command in readme.md. I will update it.

  #11 opened 3 years ago by jinx-USTC
  2

• why don't use pseudo_labeled=-1 ?

  #21 opened 3 years ago by henrysun1996
  0

• I downloaded the trained model you released, but only market and Dukle. Can you publish the model trained on VeRi?

  #18 opened 3 years ago by lbl123456
  1

• Can't reproduce MSMT17 results

  #14 opened 3 years ago by hu-zheng
  8

• Why hard sampling is only beneficial for resnet_ibn_a?

  #22 opened 3 years ago by tengxiao14
  0

• 关于use hard的问题

  #17 opened 3 years ago by xmy0916
  4

• i see in the experiment you only train in batch 256,different from that in method like spcl,and is the good result get by larger batchsize?

  #12 opened 3 years ago by linwang123456
  7

• learning with sourcedomain?

  #9 opened 4 years ago by talentCODE
  1

• Question about memory updating in q_hard

  #2 opened 4 years ago by nightwork23
  4

• Parameter configuration for Duke and MSMT17 dataset, as well as memory initialization

  #5 opened 4 years ago by Terminator8758
  1

• why the cluster_loader uses test data augmentation mode instead of training mode?

  #8 opened 4 years ago by knaffe
  1

• faiss:gpu out of memory

  #4 opened 4 years ago by CompliceU
  3

• duke on 1 GTX 2080ti,the mAP can not reach 72.8%

  #6 opened 4 years ago by Git-Yxt
  1

• question about the ablation studies

  #1 opened 4 years ago by hjwang-824
  2

• 有没有可视化的例子呢？

  #3 opened 4 years ago by opentld
  2