NVIDIA Transfer Learning Toolkit
The NVIDIA Transfer Learning Toolkit is ideal for deep learning application developers and data scientists seeking a faster and efficient deep learning training workflow for various industry verticals such as Intelligent Video Analytics (IVA) and Medical Imaging. Transfer Learning Toolkit abstracts and accelerates deep learning training by allowing developers to fine-tune NVIDIA provided pre-trained models that are domain specific instead of going through the time-consuming process of building Deep Neural Networks (DNNs) from scratch. The pre-trained models accelerate the developer’s deep learning training process and eliminate higher costs associated with large scale data collection, labeling, and training models from scratch.
The term “transfer learning” implies that you can extract learned features from an existing neural network and transfer these learned features by transferring weights from an existing neural network. The Transfer Learning Toolkit is a Python based toolkit that enables developers to take advantage of NVIDIA’s pre-trained models and offers technical capabilities for developers to add their own data to make the neural networks smarter by retraining and allowing them to adapt to the new network changes. The capabilities to simply add, prune and retrain networks improves the efficiency and accuracy for deep learning training workflow.
Enabling End to End Deep Learning Workflow for Intelligent Video Analytics
For developers designing and integrating Intelligent Video Analytics (IVA) end applications such as parking management, securing critical infrastructure, retail analytics, logistics management and access control, etc, NVIDIA provides end to end deep learning workflow with Transfer Learning Toolkit for accelerating deep learning training and deploying with DeepStream SDK 3.0 on Tesla GPUs. The models are fully trained for IVA specific reference use cases such as detection and classification. Here is a list of pre-trained models for image classification and object detection that ship with Transfer Learning Toolkit for IVA:
|Image Classification||Object Detection|
To learn more about Transfer Learning Toolkit for IVA, using pre-trained models offered for IVA and examples check out our latest blog on how to accelerate Intelligent Video Analytics applications with Transfer Learning Toolkit and DeepStream SDK for inference.
End to End Deep Learning Workflow for Medical Imaging
Transfer Learning Toolkit for Medical Imaging provides pre-trained models unique to medical imaging plus additional capabilities such as integration with AI-assisted Annotation SDK for speeding up annotation of medical images allowing developers to have access to AI assisted labeling. The 3-D Brain Tumor segmentation model developed by NVIDIA researchers won first place for Multimodal Brain Tumor Segmentation Challenge 2018. BraTS focuses on evaluation of state-of-the-art methods for segmentation of brain tumors in multimodal MRI scans. BraTS 2018 utilizes multi-institutional pre-operative MRI scans and focuses on the segmentation of intrinsically heterogeneous (in appearance, shape, and histology) brain tumors. NVIDIA Transfer Learning Toolkit for medical imaging comes with this award winning brain tumor segmentation model developed in-house by NVIDIA researchers along with additional pre-trained models for liver and lesion segmentation, spleen segmentation etc. NVIDIA’s end to end deep learning workflow for medical imaging allows developers to use Transfer Learning Toolkit for accelerating deep learning training and deploying with Clara Platform.