AI as a Service: Quick and Scalable Machine Learning POCs with No-Code/Low-Code Tools
summary
Proof of concept uses Amazon Recognition with synthetic data augmentation to detect weld spatters on motorcycle frames, scaling a tiny dataset (~100 images) up to ~10,000 and raising detection precision from ~70% to ~93% after about 10 hours of training; the POC finishes in under 20 days. A second use case automates delivery-bill OCR and ERP entry, claiming up to 90% time savings. Notes on lighting, reflectivity and quick retraining with few new samples are included.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:04Cloud and AWS partnership
- 3:10Motorcycle frame welding QA
- 6:27POC goals and constraints
- 6:49Amazon Rekognition low-code
- 8:39Data augmentation for images
- 10:30Results and timeline
- 11:14Additional resources
- 12:23Q&A
related talks
42mHow to Build Smart Applications Without a Machine Learning Background
27mAI as a Service Scaling AI with Ease and Innovation
50mAzure AI, Power new possibilities for every organization
51mCloud AI: How to Get Started Injecting AI Into Your Applications
36mAI-Powered Real-Time Quality Control
38mReal-world Machine Learning with TensorFlow and Cloud AI
56mAI War Stories: Real world learnings delivering enterprise AI solutions
27mAI for social good