search

yangze0930 / NTS-Net

This is a PyTorch implementation of the ECCV2018 paper "Learning to Navigate for Fine-grained Classification" (Ze Yang, Tiange Luo, Dong Wang, Zhiqiang Hu, Jun Gao, Liwei Wang).
MIT License
452 stars 118 forks source link

Unable to achieve the performance mentioned in the paper #1

Closed gaoyuchris closed 6 years ago

gaoyuchris commented 6 years ago

I have trained the model by the code without any change, unfortunately，I can't achieve the performance mentioned in the paper. The parameters I set are "BATCH_SIZE = 16, PROPOSAL_NUM = 6, INPUT_SIZE = (448, 448), LR = 0.001, WD = 1e-4", the only difference is that the learing rate multiplied by 0.1 after 60 epoches and 100 epoches(which is set in your code) . I have trained 143 epoch, but the acc of the test data is just 0.415? I don't know if there is any problem here？ image

gaoyuchris commented 6 years ago

I have found the problem for my stupid mistake, and got the performance mentioned in the paper, thanks for the authours

mkumar10 commented 6 years ago

What was your issue? Was it pytorch version?

metro-smiles commented 6 years ago

I was facing the same problem too, was wondering how you were able to solve it?

gaoyuchris commented 6 years ago

My pytorch version is pytorch 0.4. My stupid error is that I commented the code of "net.eval()" by mistake when evaluating the net on test data.

metro-smiles commented 6 years ago

My pytorch version is pytorch 0.4. My stupid error is that I commented the code of "net.eval()" by mistake when evaluating the net on test data.

I made the same error myself, so really appreciate you sharing this info.