Jetson Nano capability question

Robotics & Edge Computing Jetson & Embedded Systems Jetson Nano
1

I am thinking to try the Jetson Nano. I would appreciate if it could be confirmed that the following is a viable environment.

The NN is trained with Tensorflow 2.2, and have the following layers:

  • LSTM
  • Bidirectional
  • TimeDistributed
  • Concatenate
  • Dense
  • Reshape

Could I use such network to do inference in an easy way (not building custom layer) on the Jetson Nano?
if not which part is a problem?

2

You might run into problems with not enough memory for tf.

3

Thanks for the info, that seems feasible to try to limit memory usage. But you think that there will no issue with the layers? It seems that it works for some people an dB others are struggling with the layers.