Today, we are excited to announce the release of the latest version of the NVIDIA TAO Toolkit 5.0. This newest version brings openness and greater control over AI. It supports model export in open ONNX format which brings interoperability between training and inference runtime enabling model deployment on any platform .
Highlights from this release:
- Export model in standard ONNX format and deploy on any platform—GPUs, CPUs, MCUs, DLAs, etc.
- State-of-the-art vision transformers for better accuracy and robustness against image corruption and noise.
- The ability to auto-label segmentation masks with the new AI-assisted data annotation capabilities.
- New computer vision pre-trained models for optical character detection and recognition, optical inspection, and more.
- Automatic fine-tuning of hyperparameters with AutoML on new vision transformers.
Software and the models are available for free download at NGC or GitHub
To learn more:
- Read the announcement in the blog Access the Latest in Vision AI Model Development Workflows with NVIDIA TAO Toolkit 5.0
- New tech blog Improve Accuracy and Robustness of Vision AI Apps with Vision Transformers and NVIDIA TAO
- New Video Tutorials:
- Developer resources from partners