search

cvlab-epfl / LIFT

Code release for the ECCV 2016 paper
487 stars 168 forks source link

Hard mining the Descriptor network #14

Closed 13331151 closed 7 years ago

13331151 commented 7 years ago

I follow your advice of adding hard mining, and the performance of descriptor network work better than before but its still far worse compared to your released version. I wondering if there is anything I can pay attention to? Have you pre-trained the network before mining? And I extract patch with 6*sift_scale as radius for descriptor input, is it right? As for the mining rate, I double the mining rate every 5000 batches, so the mining rate become 1 at the 35000th iteration, is it the same as what you described in the paper? I really appreciate your help because I have spent so much time in this part. Faint....

kmyi commented 7 years ago

Hi, regarding mining, we do something similar to yours. however, the mining rate goes as far as to having 5 forward passes and one backward pass. IIRC.

13331151 commented 7 years ago

the mining rate goes as far as to having 5 forward passes and one backward pass

Hi, thanks for the answer. So, do you mean that you select 1285/mining_rate hard samples out of 1285 samples in 5 iterations? Thanks in advance.

kmyi commented 7 years ago

I think yes.