Autolabeling

Overview

If you have already trained your AI previously and want to label new images, create a new Deployment with the weights generated in that experiment and then click the Label button on the top right corner.

Choose the deployment

Class names of the deployment must be identical to the class names in this dataset for detected labels to be displayed.

Autolabel your image

Press the L shortcut key or click the Label button to make your previously trained AI label this image initially for you.

Fix the labels

Fix the predictions your AI's old version made in order to later train a new and better version of it. If the autolabeler made very wrong predictions just press the K shortcut key to reject them or click the X button on the top right corner.

Submit the labels

Finally, submit your changes by pressing the E shortcut key or clicking the Submit button.