Computer Vision to Secure Your Surroundings with AI/ML Solution Built Using Open Source Tools at the Edge
Talkat Open AI + Data Forum @ Open Source Summit North America 2022 · on Tuesday, 2022-06-21· at 14:55· Room 408/409 (Level 4), JW Marriott Austin, 110 E 2nd Street Austin, TX 78701·
summary
A working edge reference implementation combines OpenVINO inference engine with GStreamer/DL Streamer, an RTSP Simple Server input, Dockerized Go microservices, an Angular/Video.js UI, EdgeX Foundry for MQTT and Vault, and Postgres using TOAST declarative partitioning to store synchronized HLS video fragments plus inference metadata. It runs person/vehicle/face models from the OpenVINO model zoo, stores only clips with detections, uses TLS and Vault for secrets, was developed on i5-class hardware (acceptable but slow at scale), and focuses on reproducible open source deployment rather than benchmarking.
AI-written summary · about
chapters (28)
- 0:00Introduction
- 1:08Agenda
- 1:27Team Introduction
- 2:37What is IoT
- 3:13Global IoT Market
- 4:06Computer Vision
- 5:01Smart City
- 6:10Project Objectives
- 6:21Intel Software Ecosystem
- 8:04Background
- 9:07Architecture
- 9:49RTSP
- 11:07H264
- 11:24GStreamer
- 12:12Inference Engine
- 15:44EdgeX Foundry
- 17:30PostgreSQL
- 23:46Go Microservices
- 25:51Angular UI
- 27:43Grafana UI
- 28:21Solution Flow
- 31:04Challenges Learning
- 31:22Resource Constraints
- 32:15Data Management
- 34:27Security
- 35:30Ethical Considerations
- 38:06Questions
- 40:28Intel Edge
related talks
29mApplied Custom Vision
28mVision, Big Data and AI
21mLeveraging Python for Real-Time Image processing
40mDemocratise public agency user access to AI/ML on VAS
26mUse computer vision at the edge to improve operations with AWS Panorama
42mComputer Vision Using Go And OpenCV 3
19mAzure-enabled vision AI with NVIDIA AI Enterprise and Jetson
19mFrom concept to production: design and optimize computer vision applications on Arm