Introducing Fritz AI Model Training

Fritz AI Model Training allows developers to use datasets they’ve generated and collected to train production-ready mobile machine learning models—without any code. Model Training makes model optimization simple and straightforward, with model variants available for a variety of needs and use cases. Both Core ML and TensorFlow Lite models are created automatically, making iOS and Android development easy.

Because Model Training exists within the same workflow as both the Fritz AI Dataset Generator and Dataset Collection System, developers never have to leave Fritz AI to build and scale production-quality models.

Fritz AI Model Training is a low-touch, no-code tool for building production-ready mobile machine learning models. Here’s a summary of how it works:

* Select the type of model you want to train: e.g. pose estimation
* Choose the datasets you want to train your model on
* Select a variant that meets your need: small, fast, or accurate
* Set a training time budget
* Begin training with a button click

This functionality listed above enables ML teams and even developers without ML expertise to:

* Build production-quality, mobile-ready ML models without code.
* Easily retrain current models with newly-generated or collected data.
* Optimize model variants according to project or device requirements.
* Avoid writing complicated pre- and post-processing code.
* Streamline workflows with a unified tool for data generation, collection, and model training.
* Drop trained models directly into the Fritz SDK for faster project development.