Layman's Guide to Implementing Custom Neural Network

I have been searching forums and articles trying to find a guide to implement a pre-trained custom network on the Jetson AGX Xavier.

Is there any step-by-step guide out there? Most of what I have found is convoluted.

Details on Network and Data:
-I am implementing the neural network and dataset found in (see page 7 for NN-model)
-Please note that the data input is a (2, 128) matrix, which ideally wouldn’t be implemented as a jpeg image as with Nvidia’s built-in models.

Any thoughts are welcome.


It doesn’t seems to be TensorRT issue.
Please file your issue in below forum to get a proper support:


Wasn’t sure how to refile, so I reposted under that thread, but now I am having trouble deleting this post.