Learning Outcomes AI Vision
Learning Outcomes
1
Understand the differences between multi-class, multi-label and object detection tasks in image classification
2
Create and configure Azure Custom Vision resources for training and deploying computer vision models
3
Upload, tag and manage image datasets using the Custom Vision portal
4
Train, evaluate and interpret the performance of classification models using metrics such as precision, recall, and average precision
5
Apply best practices for dataset design, including the use of background/negative examples to improve generalisation
6
Use the Quick Test interface to validate model predictions and iterate on model improvements
7
Create object detection models using manually-drawn bounding boxes to locate and classify multiple objects within images
8
Publish trained models and interact with them programmatically using the Custom Vision Prediction API within Jupyter Notebooks
9
Integrate Custom Vision with Azure ML Studio to run notebook-based workflows and manage resources via cloud compute instances
10
Understand how Azure AutoML can be used to train segmentation models automatically, leveraging annotated image masks and performance-driven model selection
Important Note
This module focuses on practical computer vision applications using Azure Custom Vision. While we cover essential machine learning concepts related to image classification and object detection, this is primarily a hands-on guide designed to help you implement working vision models for your research projects.
If you want to learn more about Machine Learning theory and foundations, click below for the link to our comprehensive course.