A Yolov8 detection model that segments nails in images.
The model can be used as an ADetailer model (for Automatic1111 / Stable Diffusion use), or using other inference scripts to return detection bounding boxes of watermarks.
The model is entirely trained on the following dataset: Personal Projects/Nails Segmentation
A tutorial and code how to use the model can be found on this Github: https://github.com/MNeMoNiCuZ/yolov8-scripts or this CivitAI article.
Description
FAQ
Comments (4)
ty, but photography only tho :-(
Pushing it to the limit! haha
Yeah it's only trained on photos.
Now to be fair, in your image, there's almost no visible nails, so I don't know how it could detect that.
Maybe you need a finger-tip detector? :)
@mnemic i had some issues too with this image, that its b&w shouldnt be a problem right?
https://civitai.com/images/17248788
@Rvage I feel like it should not be a problem, but I can't guarantee it. The nails did look a bit wonky there though. The training data are just real / normal nails. It doesn't contain broken generation nails.
So the model could probably be further enhanced by semi-broken generations training data, and black and white training data probably also help it.