tengshaofeng / ResidualAttentionNetwork-pytorch

a pytorch code about Residual Attention Network. This code is based on two projects from
678 stars 165 forks source link

下面是我运行某epoch的结果,我想问一下:为什么分类测试精度这么低? #17

Closed gden138 closed 5 years ago

gden138 commented 5 years ago

Epoch [32/300], Iter [100/254] Loss: 0.2530 Epoch [32/300], Iter [200/254] Loss: 0.1421 the epoch takes time: 40.39500594139099 evaluate test set: Accuracy of the model on the test images: 87 % Accuracy of the model on the test images: 0.8785185185185185 Accuracy of plane : 0 % Accuracy of car : 0 % Accuracy of bird : 1 % Accuracy of cat : 0 % Accuracy of deer : 0 % Accuracy of dog : 3 % Accuracy of frog : 0 % Accuracy of horse : 0 % Accuracy of ship : 0 % Accuracy of truck : 1 %

分类测试精度这么低,还有多个类别都有对应精度?是不是我运行软件的版本有问题,我用python3.5 pytorch1.1版本。还有就是最高精度没输出是咋回事。谢谢!

13331112522 commented 5 years ago

我也是这个问题,请问如何解决的?

gden138 commented 5 years ago

因为分类数比较多,我的40多类,代码判断里面好像是类别精度大于0.01时候才统计显示,所以有些类别显示为0,其实应该是有精度的。

------------------ 原始邮件 ------------------ 发件人: "David Zhou"notifications@github.com; 发送时间: 2019年6月5日(星期三) 晚上10:49 收件人: "tengshaofeng/ResidualAttentionNetwork-pytorch"ResidualAttentionNetwork-pytorch@noreply.github.com; 抄送: "yifan"120828069@qq.com;"State change"state_change@noreply.github.com; 主题: Re: [tengshaofeng/ResidualAttentionNetwork-pytorch] 下面是我运行某epoch的结果,我想问一下:为什么分类测试精度这么低? (#17)

我也是这个问题,请问如何解决的?

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or mute the thread.

tengshaofeng commented 5 years ago

以总的准确率为准,根据你这个看来,应该是每个单独的分类的精度计算有问题,相应改下代码就好了

gden138 commented 5 years ago

嗯嗯,我大致知道了,我看评论里面已经能实现97以上的精度了,是又加入了哪些技巧呢?能不能共享一下最新结果。谢谢

------------------ 原始邮件 ------------------ 发件人: "bojohn"notifications@github.com; 发送时间: 2019年6月10日(星期一) 下午3:46 收件人: "tengshaofeng/ResidualAttentionNetwork-pytorch"ResidualAttentionNetwork-pytorch@noreply.github.com; 抄送: "yifan"120828069@qq.com;"State change"state_change@noreply.github.com; 主题: Re: [tengshaofeng/ResidualAttentionNetwork-pytorch] 下面是我运行某epoch的结果,我想问一下:为什么分类测试精度这么低? (#17)

以总的准确率为准,根据你这个看来,应该是每个单独的分类的精度计算有问题,相应改下代码就好了

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or mute the thread.

tengshaofeng commented 5 years ago

mixup会提高一个多左右百分点