CSI camera Stops working after model is loaded

I’m doing Jetson AI Fundamentals Course from NVIDIA and I’m stuck at S1E3. I’m using CSI camera. when I execute model.to(device), live camera output stops. and it doesn’t recover from that state until i restart the device. I tried to check online, seems like no one has faced a similar issue. It would be great if i can get some help here!


Do you get the output in the beginning?
Or the camera doesn’t work at all?


I get output until I execute model.to(device). After that, only the last image ll be displayed through out.

Hi @veenabs.bhat, are you using Jetson Nano 4GB or 2GB? It would seem that model.to(device) triggers PyTorch to load/initialize a bunch of stuff on the GPU and in that time your camera times out. If possible, I would recommend temporarily pausing the camera while you are running that (or waiting to start the camera until after).

You might want to also make sure that you have adequate SWAP memory mounted and that you are running your Jetson headlessly (with the desktop UI disabled and no display attached):

If you continue having issues with all these resources running at the same time, I would also recommend continuing on with the course and Hello AI World tutorial: https://github.com/dusty-nv/jetson-inference

Hi @dusty_nv , I’m using Jetson Nano 2GB. I’ll try starting the camera later on. SWAP memory is 5085.

Thankyou for sharing the resources!

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.