I want to use coreML with vision for my trained modell, but that only takes images with RGB
colorspace. I wonder how could I convert ARFrame
CVPixelBuffer
from ycbcr
to RGB
colorspace. I need CVPixelBuffer
as a return type. I read that could be easily done with metal shaders, tho I want to use SceneKit
for the rest of the project. Is there a solution without metal shaders? If not how could I use metal shaders for this while keeping SceneKit
? Thanks for your answers in advance!
Asked
Active
Viewed 474 times
1

Samu Andras
- 171
- 1
- 6
-
1I _think_ Vision may be able to do this automatically, so that would be the easiest option (just pass the CVPixelBuffer to your Vision request object). Otherwise, vImage from Accelerate framework can do this. – Matthijs Hollemans Jul 27 '18 at 11:41
-
Yes, Vision does this automatically. – rickster Jul 28 '18 at 04:36