about datasets

Sorry to bother you. Could you tell me that how to use it on Market-1501 dataset

@huyanzhuolu Thank you for your question. If you want to use it on the Market-1501 dataset, I recommend this excellent repository. And then what you to do is to replace the loss with:

expAT/config/criterions.py

Line 6 in c132eeb

class expATLoss():

Thank you very much! But which loss should be replaced? 在 2020-11-21 17:03:13，"prism" <notifications@github.com> 写道： @huyanzhuolu Thank you for your question. If you want to use it on the Market-1501 dataset, I recommend this excellent repository. And then what you to do is to replace the loss with: https://github.com/prismformore/expAT/blob/c132eeb4e67c9d8af8033c426f0df7206a326291/config/criterions.py#L6 — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

@huyanzhuolu You are welcome. You should replace the triplet loss, which is the counterpart of expAT loss.
Using expAT loss and softmax loss as loss functions work in my experiments.
BTW, don't forget to do the common space batch normalization on the feature fed to both expAT loss and softmax loss.

@huyanzhuolu I have just updated the code for training and testing on Market-1501 and DukeMTMC-reID datasets. Please take a look!

Oh Thank you!!! 在 2020-12-12 19:25:09，"prism" <notifications@github.com> 写道： @huyanzhuolu I have just updated the code for training and testing on Market-1501 and DukeMTMC-reID datasets. Please take a look! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

If you have any question, plz let me know. I will close this.