Jetson Reboots soon when using webcam and mtcnn.detect

Autonomous Machines Jetson & Embedded Systems Jetson Nano
, ,
#1

Hi,

I wanted to run some simple python scripts with face detection on jetson nano 2gb, however it causes jetson to reboot soon after the script starts. I expect the issue is around line mtcnn.detect(frame, landmarks=True) (facenet-pytorch lib). This happens in both cases when detection is running on gpu or cpu.

Can you help me figure out the cause? Is it cpu/gpu overheating? …doing out of memory? Where can I find any related logs or even better avoid reboots?

thanks

#3

Hi,

Sometimes the power starvation will cause Jetson to reboot.
To classify the root cause, could you set the nvpmodel to 5W mode and try it again?

$ sudo nvpmodel -m 1

Thanks.

#4

Hi AastaLLL,

same happens with nvpmodel -m 1 or nvpmodel -m 0 . What does this mean?

Unfortunately, I do not have direct access to the charger to check voltage etc. right now…

I also noticed that it takes ~30 sec longer before the reboot when running the computation on cpu as compared to running on gpu.

Thanks

#5

Hi,

This will setup Nano for different power budget.
You can find the detailed below:

https://docs.nvidia.com/jetson/l4t/index.html#page/Tegra%20Linux%20Driver%20Package%20Development%20Guide/power_management_nano.html#wwpID0E0FL0HA

If the issue still happens, could you share the detailed reproducing steps with us?

Thanks.

#6

Hi AastaLLL,

I finally managed to get my hands on the jetson again, Steps to reproduce the issue are:

  1. start jetson
  2. start gui sudo systemctl start gdm3.service
  3. connect webcam (quite old one actually Microsoft Lifecam VX 1000)
  4. python3 webcam.py (attached)
  5. jetson reboots after a couple of seconds

webcam.py (1.7 KB)

Same happens for nvpmodel -m 1 or 0 . My power supply output is 5V=3A/9V=2A/10V=1.5A and no reboots happens while using jetson nano for browsing web etc.

Do you have any ideas how to track down the problem?

#7

Hi,

Thanks for your feedback.

We are going to reproduce this issue in our environment.
And share more information with you later.