Open RahulRagesh opened 5 years ago
I believe so, since Densenet may require a GPU for faster inference (even more computation for backprop when training). Which is not normal for me is my demo will stuck in grasping section, when gripper grasps a object, it will stop and not moving on anymore.
I was trying to run the demo as per the instructions. Everything builds well and works fine. Just that it is very slow. I've run it for over an hour now, there are 3 more blocks yet to be picked.
I am running it on Ubuntu 16.04 LTS. Processor Intel Core i7-7700 CPU @ 3.60 Ghz x 8
Is a GPU absolutely necessary for real-time performance using pre-trained models?