Vision, Big Data and AI
summary
AvidBeam converts real time surveillance into metadata and alerts and claims it scales to tens of thousands of cameras, citing city fleets like London with ~400,000 cameras and raw video rates of ~60 GB per hour per camera. Platform uses standard CV stacks and accelerators and uses DeepStream and OpenVINO on NVIDIA/Intel GPUs, plus Kubernetes, MongoDB, Elasticsearch and Kafka for scaling. Built and deployed features include parking reservation in Riyadh, crowd, loitering, PPE and LPR detection, and won CES Best AI Startup.
AI-written summary · about
chapters (11)
- 0:00Vision, Big Data and AI
- 1:26Video analytics purpose
- 3:12Real-time data processing
- 5:37AI video analytics tools
- 7:23Video analytics benefits and challenges
- 9:31Industries using video analytics
- 10:14Video analytics use cases
- 16:41Crowd detection use case
- 18:28Computer vision with generative AI
- 21:20AI ethics in computer vision
- 27:17Future of video analytics
related talks
40mDemocratise public agency user access to AI/ML on VAS
29mApplied Custom Vision
57mDrones & AI - What's all the buzz about ?
02mWahtari's Smart City solution
46mComputer Vision to Secure Your Surroundings with AI/ML Solution Built Using Open Source Tools at the Edge
22mBuilding the AI City: Edge-to-Cloud AI Inference for Smart Cities
36mVideo analytics at scale: Deep Learning, Computer Vision, ML & Spark
35mLarge-Scale Video Intelligence