search

dusty-nv / jetson-inference

Hello AI World guide to deploying deep-learning inference networks and deep vision primitives with TensorRT and NVIDIA Jetson.
https://developer.nvidia.com/embedded/twodaystoademo
MIT License
7.76k stars 2.97k forks source link

ImageNet failed to load network #1367

Closed eanmikale closed 1 year ago

eanmikale commented 2 years ago

Dusty, recently trained a network on the Jetson Xavier AGX, pointing all data to my SD Card. I am now running the model on the Nano, using the same card for rapid prototyping. However, I am getting the following error, that the forums have not been able to help me solve:

[TRT] imageNet -- loaded 280 class info entries [TRT] imageNet -- didn't load expected number of class descriptions (280 of 123) [TRT] imageNet -- failed to load synset class descriptions (280 / 280 of 123) [TRT] imageNet -- failed to initialize. jetson.inference -- imageNet failed to load built-in network 'googlenet' Traceback (most recent call last): File "/usr/local/bin/imagenet.py", line 55, in net = jetson.inference.imageNet(opt.network, sys.argv) Exception: jetson.inference -- imageNet failed to load network

I searched through the data and network directories, as well as with the synset file and could not find anything corrupt, or erroneous.

Maybe using the last save checkpoint? I exported the file on a different machine. Does that matter. It took 3 days to train. Please don't tell me I need to retrain and immediately export? Any assistance would be great.

dusty-nv commented 2 years ago

Hi @eanmikale, what is the command you are using to run your model?

It looks like the labels.txt has more entries than expected - your labels.txt has 280 classes, while the model has 123 classes.

eanmikale commented 2 years ago

Dusty,I got it to run. I used the checkpoint as opposed to best model, which the chechpoint had proper classes. This solved my issue. Hope it helps others.  ---- On Fri, 18 Feb 2022 11:33:45 -0600 @.*** wrote ---- Hi @eanmikale, what is the command you are using to run your model? It looks like the labels.txt has more entries than expected - your labels.txt has 280 classes, while the model has 123 classes.

—Reply to this email directly, view it on GitHub, or unsubscribe.Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you were mentioned.Message ID: @.***>