AI acceleration at the edge
summary
Teams use Intel Geti to train vision models with minimal data, convert them with OpenVINO for one‑flag deployment across CPUs and GPUs, and run ai.io’s aiScout pipeline on AWS DL1/Xeon to sign 111 players using 3DAT that extracts 20+ keypoints and computes 1,400 biomechanics metrics from phone or broadcast video. Edge GPUs (Flex series) offload media; 4th‑Gen Xeon AMX claims up to 300% inference speedup while simple software tweaks yield pandas 90×, scikit‑learn 38×, and TensorFlow ~3× improvements.
AI-written summary · about
chapters (10)
- 0:00Introduction to edge AI
- 2:58Deployment architectures
- 5:09Tools for AI acceleration
- 8:02Securing edge deployments
- 9:45Computer vision applications
- 12:00Sports tech case study
- 20:47Edge media pipelines
- 23:26Rugged industrial hardware
- 32:12Far-edge device integration
- 39:41CPU acceleration and conclusion
related talks
09mTackling edge computing with Intel's AI Builders program
24mTools & Techniques for a cost optimized Analytics & AI pipeline
13mAccelerating AI algorithms
30mAI: unleashing the next wave
06mAI Inference at the Edge: Beyond the Hype
04m4 real-world AI deployments at enterprise scale
05mAccelerate AI development from cloud to client
48mThe Power of Edge AI: Revolutionizing Future of Real-Time, Autonomous Processing