Shenghua Fan-Whu

Results 14 comments of Shenghua Fan-Whu

fine....I also find this problem....The loss cannot go down even my lr is 1e-7...I do not know how to solve this case...I replace resnet with swin-s in my nets as...

希望作者看到以后可以回复一下~非常感谢!

freeze了,我把grad_require限定为只有fc更新,训练完以后,测试效果还不如不微调。请问您微调fc时,训练轮数也是和之前步骤的训练轮数都保持一致了吗?

好的,非常感谢您能回复~

I get runtimeerror:address has already inused when I follow README steps about single node multi GPU...could you please tell me how to solve it?

Hey,dude,I have solved this problem by modifying some codes,I just adjust the way of DDP launch and I start the single node training with multi gpus successfully.But here comes a...

还有,我在复现过程中发现,不需要训到1000epoch,训练到差不多600epoch时,IOU已经差不多是0.86左右了。请问您这样正常吗?

好的,非常感谢您的答疑解惑

> hello,请问你是用哪个toolkit测的呢?我建议用got10k的toolkit来验证结果。pysot的好像有点bug 我使用的是您readme里建议的pysot的toolkit测试的,好的,我试一下got10k的toolkit验证一下。非常感谢

> pysot验证uav和nfs这两个数据集都有点问题,测试是没问题的,算AUC有点问题,用pysot测的结果可以直接换got10k来验证一下 好的,非常感谢您耐心解答,我验证一下看看再来这个贴告诉您结果。