Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Deformable convolution (DCNv2) out of the box #301

Open
breznak opened this issue Jan 24, 2020 · 4 comments
Open

Deformable convolution (DCNv2) out of the box #301

breznak opened this issue Jan 24, 2020 · 4 comments

Comments

@breznak
Copy link
Contributor

breznak commented Jan 24, 2020

I've had some troubles installing Yolact++ (with DCNv2), as have others #251 etc.

Would it be possible to make DCNv2 working out-of-the box? I see pytorch/vision#1586 has recently been merged to pytorch-vision. Can we use that instead of compiling our DCNv2? (no, DCNv1 only)

Also, please see my comment
pytorch/vision#1586 (comment)
what is the difference between the 2 DCNv2 implementations?

@breznak
Copy link
Contributor Author

breznak commented Jan 24, 2020

Tracking DCNv2 support in pytorch-vision pytorch/vision#1788

@dbolya
Copy link
Owner

dbolya commented Jan 24, 2020

I hate compiling CUDA too so I'd love for pytorch to come with DCNv2 support out of the box. If they add it, I'll definitely replace the compiled version in this repo with that.

@dbolya
Copy link
Owner

dbolya commented Jan 28, 2020

Off topic, but it seems that issues #302-#305 have just ceased to exist. I think @breznak, you were #304 and #305, and @enhany was #302 and #303. Do you know what happened to them? This is kinda spooky.

@dbolya
Copy link
Owner

dbolya commented Jan 28, 2020

Nvm, #303 and #304 are pull requests, and it seems @enhany deleted their account, which also deleted #302 and #303.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants