Rekognition Live Demo
summary
Live demo of Amazon Rekognition shows building and deploying a custom image model that identifies specific animated characters (Big Buck Bunny) and returns bounding boxes, confidence scores and timestamps for video frames. Workflow uses images in S3 with automatic folder labeling, trains with tens to hundreds of images, reports per-label precision/recall/F1, and produces a hosted endpoint callable via the detect-custom-labels API. Video processing uses OpenCV to extract frames and the demo model performs as evaluated by the console metrics.
AI-written summary · about
chapters (9)
- 0:00Introduction to AI at AWS
- 2:28Amazon Rekognition overview
- 4:18Demo of Rekognition features
- 9:00Using Rekognition with Python
- 10:51Introduction to Custom Labels
- 14:04Custom Labels demo walkthrough
- 23:29Custom Labels API and video processing
- 27:20How Custom Labels works behind the scenes
- 28:42Summary of Custom Labels benefits
related talks
29mCOVID-19: Identity verification & work safety with Amazon Rekognition
01h 41mGetting started with Machine Learning on AWS.
40mThe Effortless Development of Custom Computer Vision Models
01h 06mAmazon Machine Learning to Predict Responses
01h 00mOpening Keynote
55mMachine Learning and AI in Java on AWS
39mUsing AWS Artificial Intelligence services with Python
42mProduct Recognition in Retail/E-Commerce with Azure AI