Clara Training Framework is the application level package of NVIDIA’s Clara Train SDK, a python-based SDK that allows developers looking into faster implementation of industry specific Deep Learning solutions to leverage optimized, ready-to-use, pre-trained models built in-house by NVIDIA. These pre-trained models are packaged as Medical Model Archives (MMARs) to accelerate the developer’s deep learning training process and reduce higher costs associated with large scale data collection, labeling, and training models from scratch. MMARs contain all the configurations needed to train the models as well as the scripts needed for model development tasks.
Clara offers an end-to-end workflow for accelerating Deep Learning training and inference for Medical Imaging use cases. The models provided are fully trained for Medical Imaging specific reference use cases such as organ and tumor segmentation and classification.
Pre-trained models and MMAR examples are available to download for specific classification and segmentation use cases with complete details available on NGC.