Jetson first boot and demo apps
Build it step by step
0 of 8 doneWrite the JetPack image
From NVIDIA's guide for your kit.
Boot and do first setup
Account, network, time zone.
Update the system
sudo apt update and upgrade.
Install jetson-stats
Then open jtop.
Run the classification demo
imagenet on a sample image.
Run the detection demo
detectnet on an image and then the camera.
Try the power modes
nvpmodel and note FPS and temperature.
Shut down safely
sudo shutdown now.
How do you build a jetson first boot and demo apps?
Set up a Jetson Orin Nano, run image classification and object detection demos on a camera, and watch power modes and temperature. You need NVIDIA Jetson Orin Nano Super Developer Kit, IMX219 camera for Jetson Nano and Orin, NVIDIA Jetson Orin Nano 19 V power adapter and USB A to USB-C cable (data). It takes 8 steps and about 120 minutes.
- Level
- Beginner
- Time
- About 120 min
- Parts
- 4
Parts list
Tick what you already own. It saves to My Garage, so “Build with what I have” knows.
Wiring
| From | To | Why |
|---|---|---|
| Power adapter | Jetson DC jack | Power (use the adapter for your kit) |
| Camera ribbon cable | Jetson CSI camera connector, contacts facing the board as the connector shows | Camera |
| Display | DisplayPort or an adapter, with a USB keyboard and mouse | First setup (or use SSH later) |
| microSD card or NVMe | Jetson storage slot | Operating system |
Code
A Jetson is a small computer with a powerful graphics chip, made for running neural networks at the edge. Follow NVIDIA's current getting-started guide for your exact kit: older boards may need a firmware update before they can run the newest JetPack, then you write the JetPack image to the card, boot, and complete the first-time setup with a strong password. After updating, install jetson-stats so you can watch the CPU, GPU, fan and temperature in jtop. The jetson-inference project from NVIDIA's community runs ready-made models in a container, so there is nothing to compile; the first run downloads models and builds optimised engines, which takes several minutes and then caches. Run classification on a sample image, then detection, then live on the CSI camera. Try each power mode with nvpmodel and note how frame rate, power and temperature change; higher modes run faster but hotter, so make sure the fan is spinning and the unit has air. This is a learning exercise, so compare numbers and write down what you see. Leave the board powered from the supplied adapter and shut down with sudo shutdown now before removing power, to protect the card.
# ===== File 1: terminal commands for the demos (run on the Jetson) =====
sudo apt update && sudo apt upgrade -y
sudo apt install -y python3-pip git
sudo pip3 install -U jetson-stats # adds the jtop monitor, then log out and in
git clone --recursive https://github.com/dusty-nv/jetson-inference
cd jetson-inference
docker/run.sh # opens a container with the demos, pick your model downloads
# Inside the container:
imagenet images/orange_0.jpg images/test/output.jpg # classification
detectnet images/peds_0.jpg images/test/peds_out.jpg # detection
detectnet csi://0 # live camera detection
# Back on the host, power mode and temperatures:
sudo nvpmodel -q # show the current mode
sudo nvpmodel -m 0 # switch mode (numbers depend on your model)
tegrastats # live CPU, GPU, memory and temperature
# ===== File 2: live_detect.py, the same idea in your own Python (run inside the container) =====
import jetson_inference
import jetson_utils
net = jetson_inference.detectNet("ssd-mobilenet-v2", threshold=0.5)
camera = jetson_utils.videoSource("csi://0")
display = jetson_utils.videoOutput("display://0")
while display.IsStreaming():
img = camera.Capture()
if img is None:
continue
detections = net.Detect(img)
display.Render(img)
display.SetStatus("Objects {:d} | {:.0f} FPS".format(len(detections), net.GetNetworkFPS()))Test checklist
What you should see at each check. If a reading is far off, stop and use the troubleshooting list.
| Check | Expected |
|---|---|
| Boot to the desktop | Setup completes |
| imagenet on the orange photo | Label printed with a confidence |
| detectnet on the camera | Boxes appear on people and objects |
| jtop open | GPU and temperatures update live |
| Change power mode | Frame rate and temperature change |
If it does not work
No picture
Check the display adapter and try another cable.
Camera not found
Reseat the ribbon with contacts the right way round and check the connector latch.
First demo seems stuck
Engines are being built, wait several minutes.
Frame rate low
Check the power mode and the fan.
Board hot
Improve airflow, or drop to a lower power mode.
Container fails to start
Reboot and rerun docker/run.sh.
Still stuck? Find the fault step by step with your multimeter →
Common questions
What parts do I need for the jetson first boot and demo apps?
NVIDIA Jetson Orin Nano Super Developer Kit, IMX219 camera for Jetson Nano and Orin, NVIDIA Jetson Orin Nano 19 V power adapter and USB A to USB-C cable (data).
How long does the jetson first boot and demo apps take to build?
About 120 minutes for a beginner maker, in 8 steps.
How do I know the jetson first boot and demo apps works?
Boot to the desktop: Setup completes. imagenet on the orange photo: Label printed with a confidence. detectnet on the camera: Boxes appear on people and objects.
No picture: what should I check?
Check the display adapter and try another cable.
Camera not found: what should I check?
Reseat the ribbon with contacts the right way round and check the connector latch.
First demo seems stuck: what should I check?
Engines are being built, wait several minutes.
Frame rate low: what should I check?
Check the power mode and the fan.
