You can automate eye or hand detailer but can't find anything that can automask pussy. That's why I trained this model so I can automask then inpaint with low denoise to get more detailed results.

If you are planning to add this to your workflow it's better to save the original image first then you can do automated inpaint and save the inpainted result and chose the best one.

Trained yolov8x model.

Training set: ~1,838 images (80% of 2297)

Validation set: ~459 images (20% of 2297)

• Precision 96.5%

• Recall 95.1%

• mAP50 97.1%
  

Model summary (fused): 112 layers, 68,131,272 parameters, 0 gradients, 257.4 GFLOPs

Class Images Instances Box(P R mAP50 mAP50-95):

all 459 467 0.965 0.951 0.971 0.573

pussy 459 467 0.965 0.951 0.971 0.573

Speed: 0.3ms preprocess, 5.4ms inference, 0.0ms loss, 0.6ms postprocess per image