We’re excited to share the latest release of the NVIDIA TAO version 6 — making it easier to fine-tune vision AI foundation models for your domain-specific applications.
What’s new in TAO 6
- New vision foundation backbones and task heads to fine-tune models. Includes C-RADIOv3, NV-DINOv2, RT-DETR, SegFormer, and more)
- Self-supervised learning to fine-tune with unlabeled data
- Knowledge distillation to boost inference throughput
- Fine-tuning microservices to scale training jobs, on prem or cloud
- Visual ChangeNet expands to support more than one golden image to improve change detection accuracy.
- New state of the art depth estimation models: NV-DepthAnythingv2 and Foundation Stereo
✅ Check out additional resources
Stay up to date by subscribing to our newsletter and following NVIDIA AI on LinkedIn, Instagram, X and Facebook.