In this release, we’re introducing
- A pose-estimation model that supports real-time inference on edge with 9x faster inference performance than the OpenPose model.
- PeopleSemSegNet, a semantic segmentation network for people detection.
- A variety of computer vision pretrained models in various industry use cases, such as license plate detection and recognition, heart rate monitoring, emotion recognition, facial landmarks, and more.
- CitriNet, a new speech-recognition model that is trained on various proprietary domain-specific and open-source datasets.
- A new Megatron Uncased model for Question Answering, plus many other pretrained models that support speech-to-text, named-entity recognition, punctuation, and text classification.
- Training support on AWS, GCP, and Azure.
- Out-of-the-box deployment on NVIDIA Triton and DeepStream SDK for vision AI, and NVIDIA Jarvis for conversational AI.
Download Transfer Learning Toolkit and pre-trained models ( computer vision | conversational AI )
Check out the dev news and the new pose estimation developer blog
- Dev news: Fast-Track Production AI with Pretrained Models and Transfer Learning Toolkit 3.0
- Pose estimation dev blog : Training and Optimizing a 2D Pose-Estimation Model with the NVIDIA Transfer Learning Toolkit. Part 1 | Part 2
Other developer blogs & resources from data generation and data annotation partners:
- AI Reverie: Preparing Models for Object Detection With Real and Synthetic Data and the NVIDIA TLT
- SKY ENGINE: Accelerate model development and AI training with synthetic data using SKY ENGINE AI platform and NVIDIA Transfer Learning Toolkit
- LIGHTLY: How to go from a quick prototype to a production ready object detection system using active learning and Nvidia TLT
- Hasty AI: How to get AI production-ready with Hasty.ai and NVIDIA TLT