3 ms·
Speaking of that. I’d think we could possibly train on simulations. We’d have the 3d information plus the rendered 2d camera view.
by mrfusion 8y ago
Speaking of that. I’d think we could possibly train on simulations. We’d have the 3d information plus the rendered 2d camera view.
- PAClearner 8y agoI don't know if people do this for vision yet but I know some people doing this for RL for robotic control.
- annerajb 8y agoGoogle has a simulation for their cars that does this. Nvidia has a hw suit for this purpose too and for testing autopilot hardware by feeding it data from another rack/server