Another interesting application of YOLOv8 is as an object detector to speed up your annotation workflow. Like ChatGPT to speed up the time it takes to write an email or code, YOLOv8 is perfect as a foundation for AI-assisted annotation.
Encord Annotate supports a new approach based on micromodels, which are intentionally overfit models trained on only a few labels for specific use cases.
We can divide it into the following steps:
first step:
Upload the input images you want to annotate to Encord's platform via the SDK from your cloud storage bucket (e.g. S3, Azure, GCP) or via the GUI.
Step two:
Label 20 samples of any custom object defined in your ontology (in this example, we will use aircraft from the Airbus aircraft detection dataset as an example).
third step:
Next is model training. On the platform, you navigate to the Models tab and start training of a micromodel with the YOLOv8 backbone (an overfitted object detection model).
Tip: You can also train tiny models for image segmentation and classification.
the fourth step:
Wait a few minutes while the model trains on your initial samples.