This ADetailer model helps to annotate eyes on images to inpaint them automatically. Model was trained using yolov8s model.
Usage:
Copy into your "models/adetailer" folder and extract
Activate adetailer extension and select model by it's name
I suggest to use it with 0.3 threshold and mask only the top k largest 2 and mask merge mode - merge, but usually detections outputs are with confidences 0.7 and more.