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

Gpu implementation #48

Open
edgarriba opened this issue Oct 30, 2015 · 6 comments
Open

Gpu implementation #48

edgarriba opened this issue Oct 30, 2015 · 6 comments

Comments

@edgarriba
Copy link
Contributor

Hi,

What do you think about using the GPU during the matching process or BA? That could a nice improvement in terms of speed.

@paulinus
Copy link
Member

Best fit for GPU is feature extraction and matching. For this, calling opencv's implementation is what would make more sense. There is no python API for that yet AFAIK.

BA is probably too ambitious

@edgarriba
Copy link
Contributor Author

Last week I received a couple of GPUs, I'll try to make a Python wrapper to use it in csfm.c

By now, in cudafeatures2d are only available ORB and BruteForce matching.
http://docs.opencv.org/master/d6/d1d/group__cudafeatures2d.html#gsc.tab=0

@gateway
Copy link

gateway commented Dec 6, 2016

I wanted to see if their was any updates on adding gpu support to this? See other discussion over at Open Drone Map to figure out if processing can be accelerated with GPU/s . OpenDroneMap/ODM#92 (comment)

@theomessin
Copy link

Any sort of update on this?

@5y
Copy link

5y commented Jul 5, 2019

Hello, Is there any GPU feature yet ?

@YanNoun
Copy link
Member

YanNoun commented Mar 15, 2021

A tentative GPU branch is pending for merge here : #623

Wasn't yet merged, as the speed-up wasn't dramatic compared to multithreaded CPU. More detailed comparison on > 20K features might prove it is maybe faster, but for now, multi-threaded CPU still wins.

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

No branches or pull requests

6 participants