Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about building practical edge AI solutions through this 21-minute conference talk that focuses on real-world deployment challenges where power is limited, budgets are constrained, and latency must be predictable. Discover how to move beyond theoretical inference performance to create reliable systems for retail environments, manufacturing facilities, urban infrastructure, and autonomous robotics applications.
Explore the Open Edge Platform, a comprehensive Linux-based foundation that integrates an edge microvisor toolkit, remote fleet management capabilities, and extensive library support. Understand how OpenVINO serves as the core inference engine, providing a unified API that targets CPU, integrated GPU, and NPU on Intel Core Ultra processors while maintaining compatibility across popular machine learning frameworks. Examine how video-intensive workloads leverage Deep Learning Streamer to create analytics pipelines that deliver clean data signals to dashboards through efficient messaging systems.
Dive into four specialized Edge AI Suites designed for specific industries: Retail, Manufacturing, Metro, and Robotics. See how retail implementations improve order accuracy and enhance self-checkout experiences using small Vision Language Models deployed on-premises. Learn about manufacturing solutions that combine video analytics with time series data and Human Machine Interface workflows. Discover Metro suite applications including smart search, person re-identification, parking management, lighting detection, and visual quality assurance for urban safety improvements.
Investigate the multi-sensor SceneScape service that fuses intersection data to enable autonomous route planning and incident response capabilities, all running locally with OpenVINO quantization. Understand the platform's commitment to interoperability through upstreamed PyTorch optimizations, integration pathways for llama.cpp and Ollama, Windows ML compatibility, and ARM host support. Learn about device qualification workflows and validated hardware profiles that enable scaling from four-camera configurations to systems handling sixty or more video streams, ensuring optimal platform sizing for diverse deployment scenarios.
