Allows to run classification models created with Microsoft's Lobe in vvvv.
Lobe is a software that "helps you train machine learning models with a free, easy to use tool". Via a simple GUI, you provide pictures, label them, and the software trains a model that you'll be able to use in your vvvv patch.
For now, Lobe only does images classification, but they claim Object Detection and Data Classification will be available soon.
As Lobe seems to be a very recent tool, things might change over time with this plugin.
When it has trained a model, Lobe exposes a local HTTP Server with a REST API that allows you to send an image to it and get classification result. In this plugin, the Classifier
node takes care of that. Simply give it the URL generated by Lobe and an Image
, and you'll get results.
Check the help browser, a help patch comes with the plugin.
Appart from exposing this REST API, Lobe also allows you export a TensorFlow model and provides a .NET API to run those. A documentation is provided on the lobe.NET repo (you have to convert the model to ONNX), but I did not manage to go through those steps yet. This would allow you to run a model in vvvv without having the software open, which can be nice.
Go to VL's command line and type
nuget install VL.Lobe -pre