With Dream ML you can train your device to distinguish any object. So you can experiment with artificial intelligence and at the same time create useful little tools.
Your own models
The creation is done in 3 steps. Entering a suitable name for the model, creating categories to distinguish it and collecting training/reference pictures of these categories.
Afterwards Dream ML will calculate a suitable model for you offline on your device. You can test it live directly with the camera of your device.
In the main window you see all models and can test, export, delete and create new ones.
The app contains a handful of informative articles, which make it easier to use Dream ML.
The article "First Steps" will guide you through the process of creating your first image recognition model. The article "Optimize" gives you a bunch of advice to refine your models. In addition, the article "Export" explains how to use the created models in your own projects.
Dream ML also offers you 21 successes, which you can achieve one after the other. There are 7 different kinds of achievements, each with 3 levels.
From the first creation of a model, over categories and pictures up to the export of models there are achievements to master.
About Dream ML
This application was created and implemented by Noah Bussinger. The source code was developed and written inhouse. As rendering framework Phaser JS v3 was used and ported to the native iOS system together with Apache Cordova. The machine learning functionality is provided by Apple's CoreML framework. The icons were created by Icons8.