Analyzing Traffic Video Streams at Scale Using NVIDIA AI Software and NVIDIA A100-Powered AWS Instances

Enterprises across industries are adopting AI to deliver better services, improve operations, and gain a competitive edge. One of the fastest-growing areas is computer vision (CV), which enables use cases including fraud prevention, identifying hot spots in retail stores, and robotics. Building and training CV models can be complex and time-consuming due to large training data, parameter size, and desired model accuracy. Inference on thousands of video streams at scale further adds complexity to this workflow. We'll show how to use Transfer Learning Toolkit and pre-trained models to build CV models and run inference simultaneously on over a thousand live video feeds on a single NVIDIA A100 AWS instance. We'll leverage NVIDIA AI software from the NGC Catalog on AWS Marketplace and take advantage of the Multi-Instance GPU feature of the NVIDIA A100 GPUs to build and deploy a traffic pattern analyzer. Join our interactive session as we walk through each step.