Jetson nano dev kit - yolov using CSI camera

Good day, as it’s first time I’m using the Jetson nano dev kit, I need know where I can find a going tutorial with yolo or an alternative program for use it as persons… objects tracking using an CSI arducam
Suggestions are super appreciated

Hi,

Do you use Jetson Nano or Jetson Orin Nano?

Since the tutorial is quite different.

Thanks.

Good day,I’m using Jetson nano with jetpack 4.6

Hi,

You can try Deepstream:
https://docs.nvidia.com/metropolis/deepstream/6.0.1/dev-guide/text/DS_Quickstart.html

It contains some YOLO samples by default.
Thanks.

okay but it’s tracks only object or can track also persons?

Hi,

This is an inference sample.
You can train the model for different usage with our TAO toolkit:

Thanks

Ookay thank, me I’m looking for something that I can track people and object, it’s for an experimental project.
I’m new in this kind of environment and I need also some good tutorials so I can follow for make it work

Hi,

You will need to retrain the model if the target changes.
To do this with TAO, we also have a tutorial below:

Thanks.

good day, reading a little seems that this program cannot be installed in the Jetson …maybe I’m wrong.
How about openCV? for now I was able install it with pytorch and torchvison…,but I don’t know if it can track objects coz less informations

Hi,

Yes, you will need to retrain the model on a desktop GPU.
And then copy the model to Jetson for inference.

OpenCV is preinstalled via JetPack.

Thanks.

Good day, the point its this one, I wouldn’t like streaming the video using with the object detection captures by the CSI camera using a laptop…the Jetsons will be mounted on a rover

Hi,

You can run the CSI camera with YOLO directly on the Jetson.
There are several tutorials available online.
(For example, some samples from the Ultralytics team)

But if you want to change the purpose of the model, for example, add tracking for the person.
You will need to retrain the model and this requires a dGPU environment.

Thanks.