Raspberry Pi Pico weather display (MicroPython)
Your first MicroPython project: temperature and humidity on an OLED.
Shop by category
Shop all componentsThe Jetson Orin Nano has a GPU, so it can run a neural network on live video with no cloud service. NVIDIA's open jetson-inference project ships a container with ready detectors such as SSD-MobileNet. You point its detectnet tool at a USB or CSI camera and it draws boxes and labels on every frame. This is the standard first project for any Jetson carrier board or camera.
Point cameras only at your own property and tell people that it is there.
Heat sinks and carrier boards get hot: do not touch them while running.
Tick anything you already have and the total updates.
Jetson Orin Nano Super developer kit with a power supply and carrier board
microSD card for the JetPack image (or boot from NVMe)
Follow the colours: red is power, dark grey is ground, blue is a signal.
Use the power supply that comes with the developer kit. The board draws more than a Pi; do not use a phone charger.
The container needs a display attached to the Jetson, or use `detectnet /dev/video0 rtp://<pc-ip>:1234` to stream to another computer.
The first run builds a TensorRT engine for the model. Later runs start in seconds.
Check the cable and try `v4l2-ctl --list-devices`. Another USB port can help.
Only to download JetPack, the container and the models once. Detection then runs offline.
Yes. The same project has a training tutorial to retrain a model on your own pictures.
Yes. The steps are the same for other JetPack boards, though speed varies with the model.
4 parts: NVIDIA Jetson Orin Nano Super Developer Kit, Arducam UB0238 5MP Auto Focus Mini USB Camera Board, microSD Memory Card (Class 10) and Arducam UB0240 IMX179 8MP Autofocus USB Camera Module. On this page you can tick the parts you already have and add only the rest to your cart.
It is a advanced project and takes about 3 hours. The build on this page has 6 steps.
About ₹65,687 for the parts at the prices on this page. Quantity price breaks lower the cost when you build several.
Flashing JetPack and using the NVIDIA container; Running a trained detector on a camera; Reading detections in Python.
Point cameras only at your own property and tell people that it is there. Heat sinks and carrier boards get hot: do not touch them while running.
Ideas that reuse this build: Pi USB camera motion snapshots. They are listed under Take it further on this page.
Your first MicroPython project: temperature and humidity on an OLED.
A web page with ON and OFF buttons that switches a relay over Wi-Fi.
A warm ring light that fades up at dusk and down at dawn.
Turn a Pi and a touchscreen into a home command centre with a clock and to-do list.
Learning project. Check every circuit and the datasheet of each part before you power it up.