LucasSwitz
commented
9 years ago As an update for testing results:
Pixy: Has advantages in that we don't have to write code for it. With this said, it also means we cannot edit its method of processing past basic brightness, saturation, hue, etc. Object tracking works well, but is inconsistent. Using it for practical FRC reasons would require continuous calibration and possibly unreliable results.
PiCam: Allows us to work in a familiar Linux environment. Has full support for openCV's python libraries(and C). This allows us to better manipulate the actual process of vision processing(as opposed to the Pixy) but at a much lesser quality. I was able to get about 1.5-2 frames per second by using the picam to capture the image stream and then pass that stream to openCV to be processed into a threshold image. Further testing can be done in terms of using different methods to capture and render the photos; although, I believe that no matter which way we do this the pi will not supply adequate power to give us quality vision. Conclusion: PiCamera can be used if we can create a functional vision system with very low FPS.
Jetson: Testing has not been done yet.
Kinect: Testing has not been done yet.
eddie-on-gh
The current date set for this is January 17th, 2015. I realize that this is very little time but in order to fully flesh out the necessary features, we'll need to have our mind made up fairly soon. If possible, update this Issue with information pertinent to the decision such as testing results.