Zauberzeug Learning Loop

The Zauberzeug Learning Loop provides the complete workflow for efficiently building your own specialized AI computer vision models. Images can be annotated and AI training started via an easy-to-use web interface.

But that's not all. Because the "Learning Loop" is based on Transfer Learning and Active Learning to continuously improve the models: New images from the user's own use case can be continuously uploaded via a simple API. These are automatically sorted and prioritized so that images with poor recognition results are selected for the next round of annotation. After annotation, the new data flows into the next training and can be conveniently installed on the production systems from the web interface. By iteratively repeating this process with only 20-100 images at a time, the total amount of data processed can be reduced quite significantly. This is because only the images that further improve the model need to be manually annotated. Thus, "fleet learning" is possible, as practiced by Tesla.

Thanks to the intuitive Kanban-oriented user interface, the whole process can be performed even without expert AI knowledge. Thus, the Data Scientists/Engineers can take care of the AI network architecture development in a decoupled manner, for example, and connect it to the Learning Loop via APIs.

Displayed price is the monthly licence price.