zlmzju / part_reid

Code for ICCV2017 paper: Deeply-Learned Part-Aligned Representations for Person Re-Identification

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

reproduce results

yang502 opened this issue · comments

I am try to reproduce the results on market1501. But the triplet loss is very small, nearly to zero. So after iter=35000+, the network did not update(loss=0). I can't reproduce the results(R@1:81%,mAP:63%), but get lower results (R@1:76%,mAP:55%).
Have you met this situation? What's the possible reasons ?

@zlmzju hi ,I met the same problem. my model can't reproduce the 81% result on your market.caffemodel, instead is 76%. can you explain that? I followed the same instructions you posted on docs. Thank you