The Image Tag Prediction Assistant is our offering to automate adding metadata to images. We do so by training an AI model based on your provided data. This custom-tailored approach means that you get AI assistance that improves over time as you provide more and more data.
To get started with the assistant you should:
- Add tags to your project’s taxonomy;
- Label 10 images with tags and set their status as "Done" or "To review".
This will trigger our model service to train the first iteration of the Assistant. This might take 2-10 minutes, depending on the load and the images themselves.
When it becomes available, you will receive a notification in the bottom left corner of your screen.
You can toggle the predictions on and off by pressing the toggle at the right panel of the Annotation Environment in the Image tags section.
Please note that the model will still retrain even if you have switched it off. We will just hide suggestions for you.
To see this assistant in action, please check out the Image Classification solution page on our website.
To get a deep overview of the Image Tag Prediction task please refer to our comprehensive guide
To see the predictions you should turn on the Assistant and then select an image from the bottom bar of the Annotation Environment UI. As soon as you do it, we will populate the attribute input(s) with suggestions - as we do here.
In the example above, we get some nice tag predictions for the lovely skiing picture. You can accept each suggestion individually by clicking on the plus button. Or, if you have more than one suggestion, you can accept all of them by clicking "Accept all".