Partial-Person-ReID icon indicating copy to clipboard operation
Partial-Person-ReID copied to clipboard

Published 20 hours ago • verified publisher icon JDAI-CV

The results were too different

Open xiaoweihappy123 opened this issue 6 years ago • 6 comments

Hello, could you please provide a detailed command line? The result of my running on two partial datasets is only over 50, which is far different from the result on GitHub.

xiaoweihappy123 avatar Oct 02 '18 07:10 xiaoweihappy123

@xiaoweihappy123 Me too.My result is Single Query: [cmc1: 45.07%], [cmc5: 66.80%], [cmc10: 76.07%] mAP = 0.50221015378 Done, 3.83s I guess if my mAP calculation method is wrong.I wonder what is the latest result of your code and how is it solved?Looking forward to your reply!Thanks!

xjhjinhui avatar Jan 11 '19 03:01 xjhjinhui

Search the author's latest paper for things that need to change.

xiaoweihappy123 avatar Jan 11 '19 06:01 xiaoweihappy123

@xiaoweihappy123 Could you please tell me the paper's title?Thanks a lot!

xjhjinhui avatar Jan 15 '19 09:01 xjhjinhui

recognizing partial biometric patterns

xiaoweihappy123 avatar Jan 16 '19 01:01 xiaoweihappy123

@xjhjinhui @xiaoweihappy123 I also only got 49.73 rank1 accuracy on Parital-reID dataset and 57.14 rank1 on Partial-iLIDS. Have you ever explored any setting or solution to get the performance that the author reported in paper and readme file. looking forward to your reply! Thanks!

hh23333 avatar Apr 17 '19 02:04 hh23333

Did you ever try to set spatial_train True in command line for improving perfromance?

command line is just like this: python script/experiment/train.py --dataset market1501 --partial_dataset Partial_REID --spatial_train True --total_epochs 400 @hh23333 @xiaoweihappy123 @xjhjinhui

CodeToPoem avatar Oct 22 '19 09:10 CodeToPoem