I have got a Nvidia Jetson Nano. I ran many neural network models, different types of computer vision algorithms on it, but now I need an embedded board with better computational capacity. I am planning to get a Jetson AGX Xavier and embed it on a UAV ( quadrotor). I am going to run several neural networks on it.
I have multiple questions about performance of AGX Xavier model.
1-) In Jetson Nano I got better FPS (frame per second) with CSI camera instead of USB camera which is normal. I want to ask that which CSI camera you guys suggest to me? I do not want to use hobby cameras like Raspberry Pi camera. I need something more professional with CSI spec. How about this one?
2-) As I said before, I will run multiple NNs on it. I would need an external accelerator board. Do you guys have any suggestion? Or, you can suggest a different type of Jetson series instead of Xavier?
3-) I can get more FPS by optimizing my scripts, algorithms.But, how can I get more FPS by tuning hardware (for computer vision, deep learning purpose)?