Serverless Architecture for Product Defect Detection
summary
Shows a working serverless pipeline that uses Amazon Lookout for Vision with S3, API Gateway, Lambda, Step Functions, DynamoDB, Kinesis Firehose, SNS, QuickSight and CloudWatch to do defect detection, alerting and dashboards. Claims models can start from as few as 30 images and the demo runs on a 6000+ image metal casting dataset, produces high precision/recall, sends email alerts on anomalies and writes inference records to a data lake, and the one‑click CloudFormation deploy is shown to function end to end.
AI-written summary · about
chapters (8)
- 0:00Serverless defect detection intro
- 3:54Quality in manufacturing
- 7:27Manual vs automated inspection
- 10:17Amazon Lookout for Vision setup
- 13:49Model feedback improvements
- 15:56Solution architecture
- 20:29Live defect detection demo
- 24:21Resources and next steps
related talks
38mAnomaly detection in the cloud: how to detect issues in data [...]
29mAmazon Lookout for Vision
50mDeveloping Highly Scalable Image Storage Solution with AWS Serverless
28mAutomation Engineering with Serverless Compute
49mServerless Architectural Patterns and Best Practices
50mAutomated Serverless Security Testing: Delivering secure apps continuously
51mTaking Serverless to the Next Level
41mLive Coding on Observability for Serverless