![]() |
SmartSpectra C++ SDK
Measure human vitals from video with SmartSpectra C++ SDK.
|
Directories | |
| preprocessing | |
Files | |
| preprocessing_foreground_container.hpp | |
| preprocessing_foreground_container_impl.hpp | |
| preprocessing_stream_and_packet_names.hpp | |
This is a private/internal set of tools for experimentation with and testing of Physiology Edge and the Smart Spectra C++ SDK. Smart Spectra C++ Lab does NOT get published in the public GitHub SmartSpectra repository.
The preprocessing app & foreground container are here to provide exclusive access to just the preprocessing portion of the Edge graph. Currently, it only works with the JSON version of preprocessing. It takes video, frame folder, or webcam as input, then outputs JSONs to disk in either spot or continuous modes.
Mostly follow the monorepo Build Instructions with one hugely-important addition: pass in -DBUILD_LAB_SUPPORT=ON to cmake during the configuration step. Otherwise, the preprocessing graphs will not be installed as part of the Physiology Edge dependency – and these are the ones Lab uses.
After building you can locally run this command, just replace cmake-build-debug with the name of your build folder and <YOUR_API_KEY_HERE> with your API key.
Use --operation_mode=continuous or --operation_mode=spot, depending on your needs.
Use --enable_phasic_bp to enable the blood pressure inference model if needed (will take a long wile to calibrate, though, so don't expect BP output right away).
For documentation on all command-line parameters, run
Examples:
On Debian-based Linux, if C++ SDK is built within the monorepo (e.g. via docker shell) and packaged using cpack -G DEB, like it currently happens in CI on commit to any trunk (dev, test, main), instead of building, you can grab the package from s3 and instal it:
Then, it's even easier to run: