Do I Know You? Real time facial recognition with an Apache stack
summary
Full stack lambda-style system computes Mahout/Spark distributed SVD to produce eigenfaces (130‑d vectors) from a 10,000-face dataset in about 45 minutes, uses OpenCV Haar cascade for face detection, and runs an Apache Flink online pipeline that applies Flink windowed canopy clustering (5–10s windows, drop clusters <3 to filter ghosts) then issues a Solr nearest‑neighbor search to match or append new faces. Prototype runs on cheap drones and works end‑to‑end, though image buffering to Kafka causes backpressure and the live demo had intermittent glitches.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:34System overview
- 2:42Facial detection
- 4:21Facial recognition methods
- 8:48Addressing system challenges
- 11:00Canopy clustering
- 12:51Architecture and performance
- 15:50AI and streaming contexts
- 18:22The future of open source AI
- 22:19Q&A session
related talks
45mReal-time, Streaming Advanced Analytics....
01h 12mBuilding a live face recognition system
39mBuilding Highly Scalable Facial Recognition Pipelines
53mNodeJS, ML, K8s and Unethical Face Recognition
55mMachine Learning and AI in Java on AWS
21mBring AI-Based Search to Your Web App
40m"Introduction to Face Processing with Computer Vision|"
49mReal Time Face Detection through a Webcam Using AI