Hi,
I’ll make a real-time object detection system. I’ll buy a jetson TX1 for this. I do not have a computer with an Nvidia display card. Is the Jetson TX1 alone enough for this job? Can I train the network on Jetson TX1? If you can train, which libraries you need for this training.
Thank you…
Training is usually on a full-sized desktop unit or faster. I’ve seen references that the TX1 can do training, but is slow. Additionally, there may be some software for training which does not have the arm64/aarch64 version available (most of what is useful for taking advantage of using an existing trained cuDNN runs on the TX1). Someone else will have to comment on libraries and actual experiences with training under a TX1.
I understand. Thank you so much…