这个代码可以用来分割多类吗？
Kaido0 opened this issue · 5 comments
Kaido0 commented
这个代码可以用来分割多类吗？
zizhaozhang commented
可以。需要将conv10改为输出多个channels（类的个数）。然后用softmax + cross_entropy loss
zhcv commented
我多跑分类不行啊，标签怎么制作呢
zizhaozhang commented
set opt.num_class to the one you want. The label is a HxW map where each pixel has value [0,...,#of class).
zhcv commented
I have tried , it does not work , did you run multi label segment with this repository?
zizhaozhang commented
Probably you can show me the error to let me check. It should work with
very minor modifications.
…On Wed, Nov 22, 2017 at 9:35 AM, zhapeng ***@***.***> wrote:
I have tried , it does not work , did you run multi label segment with
this repository?
—
You are receiving this because you modified the open/close state.
Reply to this email directly, view it on GitHub
<#2 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AFfrUe775gq3VN7_KdzkC-9WE3zOjoSzks5s5DFMgaJpZM4NRijV>
.
--
Best Regards,
Zizhao