Python user space driver for MobIR AIR Thermal camera

This is a user space driver for the USB-C Guide MobIR Air. It uses libusb and v4l2loopback to realize this.

Disclaimer: Most of the algorithms here have been reverse engineered from the MobIR Android app.

I'm an EU citizen and this is a tool to increase compatibility with the mentioned thermal camera

Usage

First step is to create a v4l2 loopback device, this can be done during module loading or with the v4l2loopback-ctl command.

# enable v4l2loopback with one device
modprobe v4l2loopback devices=1

The next step is to configure your loopback video device (/dev/video*) in the app, and install all dependencies (see requirements.txt).

python3 -m pip install -r requirements.txt

Afterwards one can start the driver with ./src/main.py -l /dev/videoX. It needs to be noted here, that the script requires the thermal camera to be already connected to the host device.

The /dev/videoX feed is a Y16 RAW feed, with a single pixel being Kelvin values * 100.

Notes

Note

See the "notes" branch for my WIP write-up on this project.

1. A cold start of the camera requires about a minute until the temperature stabilizes consistently. Before that point the measured temperature will fall rapidly into impossible ranges (e.g. negative °C) and return into a feasible range during each calibration.

2. The code is not guaranteed to be bug-free. For one, it does not handle edge error cases correctly, but will instead crash. On the other the resulting temperatures could very well be completely off. The algorithms in MobirAirTempUtils have been tested against the original app and seem to be correct. But this is not guaranteed for the parameters needed in these formulas.