NVIDIA Tools for Training and Deploying Intelligent Vision Applications at the Edge

Explore NVIDIA's tools for training, building, and deploying intelligent vision applications at the edge in this comprehensive one-hour video. Discover how to process data from sensors, cameras, and IoT devices, train with large datasets, and deploy real-time, high-throughput, low-latency video analytics pipelines. Learn to optimize training workflows and leverage pre-trained models for applications like smart parking, infrastructure monitoring, disaster relief, retail analytics, and logistics. Gain insights into NVIDIA's suite of tools for creating, building, and deploying video apps that deliver business efficiency. Explore topics such as transfer learning, the NVIDIA Transfer Learning Toolkit (TLT), achieving accuracy with small datasets, model pruning, purpose-built pre-trained networks, DeepStream software stack, and end-to-end deep learning workflows. Understand the challenges of video analytics, IVA application workflows, and at-scale deployment strategies, including bi-directional IoT communication and OTA model updates. Access developer resources and join the NVIDIA Developer Community to further enhance your skills in building intelligent vision applications.

Intro
CHALLENGES WITH VIDEO ANALYTICS
CREATE AI - TRANSFER LEARNING
NVIDIA TRANSFER LEARNING TOOLKIT (TLT)
TLT WORKFLOW
ACHIEVING ACCURACY WITH SMALL DATASET
MODEL PRUNING
TLT NETWORK ROADMAP
PURPOSE BUILT PRE-TRAINED NETWORKS
MODEL PERFORMANCE
IVA APPLICATION WORKFLOW
DEEPSTREAM SOFTWARE STACK
DEEPSTREAM - MANY INDUSTRIES, FLEXIBLE DEPLOYMENT
DEEPSTREAM GRAPH ARCHITECTURE
DEEPSTREAM ROADMAP
END-TO-END DEEP LEARNING WORKFLOW
PYTHON SUPPORT
DEEPSTREAM WITH TRITON INFERENCE SERVER
AT-SCALE DEPLOYMENT
BI-DIRECTIONAL IOT COMMUNICATION
OTA MODEL UPDATE
SMART RECORD
SECURITY
SUMMARY
GET STARTED TODAY
DEVELOPER RESOURCES