Introduction
As an ornithologist, you study the behavior, physiology, and conservation of birds and their habitat. Your work often involves surveying, recording, and reporting on bird activity. To help gather data, you want to build a machine learning model that identifies the species in an image of a bird. You also want to better document endangered birds species to help populations increase. Learning more about birds also is a great way to educate yourself and others about the natural phenomena in the area where you live.
In this Microsoft Learn module, you'll use the Azure AI Custom Vision service to create a machine learning model that identifies species in images of birds. We'll use the NABirds dataset from the Cornell Lab of Ornithology (Cornell Lab) to train a model to recognize species in a new photo of a bird. With new data, you can use the model to help document trends and patterns of bird habits.
The data we use to build and train our machine learning model is provided by the Cornell Lab, with a special thanks to photographers, contributors, and visitors to All About Birds. This material is based on work that's supported by the National Science Foundation under grant #1010818. This module includes a subset of the full dataset. You can download the full dataset. For specific details about the dataset, you can download a PDF file from the Computer Vision Foundation.
Learning objectives
In this module, you will:
- Get an introduction to machine learning
- Learn how to use pre-trained machine learning models in Azure AI services
- Learn how to use the Custom Vision service in Azure
- Build a custom machine learning model
- Deploy the model created by using Custom Vision
Prerequisites
- An Azure account
- A basic understanding of how to create resources in Azure
- (Optional) If you choose to use Python to upload and tag images rather than the Azure portal, a basic understanding of working with Python is required