michuanhaohao / reid-strong-baseline

Bag of Tricks and A Strong Baseline for Deep Person Re-identification
http://openaccess.thecvf.com/content_CVPRW_2019/papers/TRMTMCT/Luo_Bag_of_Tricks_and_a_Strong_Baseline_for_Deep_Person_CVPRW_2019_paper.pdf
MIT License
2.24k stars 573 forks source link

Rank1 is low #224

Open QSQSQSQS opened 9 months ago

QSQSQSQS commented 9 months ago

I use Market1501 to run the code. I set 300 epoches and the model can be trained. However, the rank1 score is rather low. The evaluation results are like this: reid_baseline.inference INFO: Validation Results reid_baseline.inference INFO: mAP: 55.4% reid_baseline.inference INFO: CMC curve, Rank-1 :2.1% reid_baseline.inference INFO: CMC curve, Rank-5 :92.5% reid_baseline.inference INFO: CMC curve, Rank-10 :97.6% The code seems fine. I can't find where the problem is.