The new Vision Portal works well for AprilTags and Tensorflow, but can the frames be passed to other tools for processing?
That was definitely the goal behind the design of the VisionPortal interface. Currently there are no other tools available with VisionPortal interfaces yet (to my knowledge).
It would be nice to be able to grab a frame and do some analysis. I assume VisionPortal is built on OpenCV?
VisionPortal uses EasyOpenCV, yes. If you’re a Java user, EasyOpenCV Documentation shows how to use EasyOpenCV, but not within the VisionPortal architecture. Unfortunately because of the architecture of OpenCV it is not directly compatible with Blocks.
The LearnJavaForFTC book (https://github.com/alan412/LearnJavaForFTC/blob/master/LearnJavaForFTC.pdf) has an example of using custom OpenCV processing with VisionPortal.
That’s helpful. Thanks!