the result compared with original paper

[YuShi1213]

I find the cross view result of this code is only 71%, lower than 88% of the original paper. Do you know why?

[BoNiu-HZY]

I use one GPU to run the train.py, it took more than one day that I didn't get the result. How long you used to run the result?

[BoNiu-HZY]

@YuShi1213

[YuShi1213]

@YuShi1213

If you run one day but didn't get any checkpoint, you should check if your dataset is read correctly. I find that if the path of your dataset is wrong, this code will not remind and can not get any result.

[YuShi1213]

@YuShi1213

If you run one day but didn't get any checkpoint, you should check if your dataset is read correctly. I find that if the path of your dataset is wrong, this code will not remind and can not get any result.

[YuShi1213]

@YuShi1213

If you run one day but didn't get any checkpoint, you should check if your dataset is read correctly. I find that if the path of your dataset is wrong, this code will not remind and can not get any result.

[BoNiu-HZY]

Thank you, I will check it.

[BoNiu-HZY]

Really appreciate your help, you are right. @YuShi1213

[YuShi1213]

Really appreciate your help, you are right. @YuShi1213

You are welcome