jisc logo

Automated Image Classification

Classify images with auto machine learning

ExploreAIVision Demos
Automated Image Classification

The following demostration showcases automated image classification using Azure Custom Vision. The model was trained to classify images of bees, wasps, other insects and other non insects based on training data, through a simple automatic machine learning pipeline.

  1. Select an example image by clicking one of the buttons below, or upload your own image.
  2. Check the classification results from the providers. Each provider shows the confidence (%) of each label.
More information about this demo

Machine learning can be tough, if you have ever had to train a model, you will know it can be difficult and time consuming to optimise and get a good model. Automated machine learning takes the hassle out of machine learning and can provide organisations without machine learning expertise, a way to utilise the predictions of machine learning models. Cloud service providers are providing a range of automatic pipelines to solve all types of problems such as image, text/document and tabular data classification and tabular data regression.

Things to consider

Uses of image classfication include tagging and categorising images. Also it could potentially be used to identify correctly constructed objects for practical courses.

Automatic machine learning can be great and the models can be updated with new training samples with time. There are some drawbacks to automatic machine learning mainly that the model is essentially a black box and you can not determine how or why the model works currently.