Hi,
I’ve been working on getting segmentation data from Webots for our vision system but have reached a problem. The cameras do not seem to be able to detect anything past the netting of the goals, as seen in the below image.
I tried setting occlusion to FALSE since it sounded like what I want but there was no change.
Is it possible to detect objects through transparent or semi-transparent objects in Webots?
I believe you are not allowed to use camera Recognition for Robocup. In any case, the TCP/IP protocol will provide you only with RGB images, nothing else. So, you will have to recognize objects from these RGB images.
That’s fine, this isn’t for direct use in the competition, just for getting data to train our network.
I see. In that case, I would suggest you to simply remove the net from the goal to work around the problem.
