"Out of Memory" prompt when detecting in real time

[agbajeayomipo287]

i have 3.3G available and i’m detecting in real time, i keep getting a “Low memory” warning after detecting for about 20 minutes. What could the issue be?
I have confirmed that it runs on the gpu.
I have a jetson orin nano.

PS: I intend to detect in real time for hours.

[kayccc]

Which JetPack SW?
Any log?
You may need to provide more details of how you’re detecting in real time on GPU, then we can have some suggestions.

[agbajeayomipo287]

import ultralytics
from ultralytics import YOLO

model = YOLO("yolov8n.pt")

Dependencies:
* torch==2.1.0a0+41361538.nv23.6
* ultralytics==8.2.38

Now, the code works perfectly well, it makes detection in real time but afyer about 20 minutes of real time detection, i get a "Low memory" prompt

After raising a discussion on the official ultralytics repo,i was asked to export to this format(.engine), although when i do so with the code below:


model.export(format = "engine", workspace = 2, int8 = True)

Error Message after running this:
[06/26/2024-12:24:43] [TRT] [W] Tactic Device request: 150MB Available: 149MB. Device memory is insufficient to use tactic.
[06/26/2024-12:24:43] [TRT] [W] Skipping tactic 2 due to insufficient memory on requested size of 150 detected for tactic 0x000000000000001e.

Thank You.`Preformatted text`

[agbajeayomipo287]

Jetpack version is 5.1.3

Forgot to add this earlier

[AastaLLL]

Hi,

Could you run the tegrastats concurrently and check the memory usage?
If the memory keeps going up during the detection, the app might have some memory leakage.

In such a case, please double-check the implementation.
Maybe you can try valgrind to help you catch the leakage.

Thanks.