TensorFlow Lite Model Maker Aims to Make Machine Learning Model Customization a Breeze

TensorFlow Lite, the framework for inference workloads on low-resource edge devices, now has a means for the customization of machine learning models: the TensorFlow Lite Model Maker.

"When bringing state-of-the-art research models to TensorFlow Lite, we also want to make it easier for you to customize these models to your own use cases," explains developer advocate Khanh LeViet of the new feature. "We are excited to announce TensorFlow Lite Model Maker, an easy-to-use tool to adapt state-of-the-art machine learning models to your dataset with transfer learning. It wraps the complex machine learning concepts with an intuitive API, so that everyone can get started without any machine learning expertise. You can train a state-of-the-art image classification with only 4 lines of code.

"Model Maker supports many state-of-the-art models that are available on TensorFlow Hub, including the EfficientNet-Lite models. If you want to get higher accuracy, you can switch to a different model architecture by changing just one line of code while keeping the rest of your training pipeline."

At launch, the TensorFlow Lite Model Maker supports two use cases: Image classification and text classification, for which tutorials are available to get people started. LeViet has indicated that work is progressing on additional use case support, including computer vision and natural language processing (NLP) — but has not offered a timescale for their availability beyond the fact they are "coming soon."

TensorFlow Lite Model Maker is available in the TensorFlow Examples GitHub repository, while the image classification and text classification tutorials are available on the Google Research website.