Custom Vision

Custom Vision AI Services

Using Custom Vision to Build Intelligent Image Recognition Models

Azure Custom Vision, from Azure's AI Services, is a powerful computer vision service that enables you to build, deploy and improve your own image classifiers and object detectors. With minimal machine learning expertise required, you can create sophisticated models tailored to your specific research needs using both Custom Vision's web-based interfaces, and more programmatic approaches by training a model from your Integrated Developer Environment.

In this module, you'll learn to implement multi-class classification for medical image analysis, multi-label classification and object detection using AI-generated image datasets, and practise advanced programmatic control using Python APIs. We'll demonstrate practical applications - from cancer detection in histological images to automated fruit recognition - giving you hands-on experience with the Custom Vision workflow.

Skills summary

Multi-class/label Classification & Object Detection

Master both multi-class/label classification and object detection techniques on a range of datasets, including medical, histiological cancer cells images.

Python API Integration

Develop programmatic control over Custom Vision models using Python in Azure Machine Learning Studio, enabling advanced customization, automation, and seamless integration into research workflows

AutoML for Computer Vision

Understand the capabilities and applications of Automated Machine Learning for image analysis, exploring advanced techniques for when standard Custom Vision approaches reach their limits