Computer Vision IRL: From Idea to On-Premise Deployment
summary
A real-world case study from Immobiliare.it on building an on-premise computer vision service to classify rooms in real estate photos. It covers data preparation, dataset creation, model development and optimization for deep learning, then deploying a REST API using FastAPI with TorchServe for model serving to handle high request throughput and sub-200ms latencies, plus monitoring of hardware, data quality, model accuracy and business KPIs using open-source tools.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 3:59Idea to production
- 7:07Building the dataset
- 9:17Model development tips
- 11:46Packaging the model
- 15:59Deployment and scaling
- 18:32Monitoring infrastructure
- 21:04Data quality checks
- 22:09Model performance tracking
- 23:36Business impact
- 24:18Key takeaways
- 25:01Q&A
related talks
01h 33mComputer Vision in Azure
20mComputer Vision: From Facial Recognition to Image Analysis
29mDeveloping Visual AI Solutions for Online Marketplace
23mComputer Vision Toolbox - Connected Computer Vision & AI Quality Inspection at Scale!
41mComputer Vision for Remote AR
29mApplied Custom Vision
27mComputer vision for automated quality inspection
46mComputer Vision to Secure Your Surroundings with AI/ML Solution Built Using Open Source Tools at the Edge