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

Using code in video processing,how can I speed up it? #74

Open
AiNi-KuangBiao opened this issue Jan 11, 2017 · 2 comments
Open

Using code in video processing,how can I speed up it? #74

AiNi-KuangBiao opened this issue Jan 11, 2017 · 2 comments

Comments

@AiNi-KuangBiao
Copy link

Dear uricamic,
I am a beginner of OpenCV . I want to use this code in real time video processing,but i find this project is no as fast as previous flandmark project .Function cvImgToCImg and detectMultiScale in video_inpuit.cpp take a long time. How can I achieve the speed show in page http://cmp.felk.cvut.cz/~uricamic/clandmark/.
Sorry for my poor english.
Many thanks.

@uricamic
Copy link
Owner

Dear @AiNi-KuangBiao,

I am not aware of such problems. Actually, the newer version is even faster (if the featurePool is used), compared to the old library flandmark, there was a significant speedup (from 10ms to 2ms on the 8 landmarks model).

Just a quick check, do you compile and use the release configuration?

Please show me some screenshot, with your processing times, or describe the process how you figured out that cvImgToCImg function is the bottleneck. Btw, if really this function is a bottleneck, you can switch to some different different image processing library than OpenCV, CLandmark core is completely independent of it, it is used just in C++ examples...

@AiNi-KuangBiao
Copy link
Author

Dear @uricamic ,
I am sorry,I compile it in debug model. I will try out it in release model again.
Thank you very much for your reply.
clandmark

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