Put the .pt model file under models/adetailer
v2-segmentation version:
An aDetailer model that should work for realistic, semi realistic, anime, small butts, large butts, humongous butts, butts having intercourse, butts wearing pants, thongs, panties, etc.
This version is trained on YOLOv8-seg - the instance segmentation model. Basically it is more specific to the part it is targeting instead of a bounding box, it is also trained on 10 additional images than v1 in hopes to help with wet/shiny butts from different angles.
Trained on 310 images, 50 epochs. It will definitely give false positives and sometimes not detect one or two but overall pretty consistent - this is technically the 9th iteration.
The example images are merely showcasing the ability to detect the ass, it is technically prior to the aDetailer being used on it. When using it on a tall image I suggest using a separate width/height to either be square or wider
v1-bbox release notes:
*thanks to anyone who participated in the early access*
Put the .pt model file under models/adetailer
An aDetailer model that should work for realistic, semi realistic, anime, small butts, large butts, humongous butts, butts having intercourse, butts wearing pants, thongs, panties, etc.
Trained on around 300 images, 50 epochs. It will definitely give false positives and sometimes not detect one or two but overall pretty consistent - this is technically the 7th iteration.
The example images are merely showcasing the ability to detect the ass, it is technically prior to the aDetailer being used on it. When using it on a tall image I suggest using a separate width/height to either be square or wider
Description
An aDetailer model that should work for realistic, semi realistic, anime, small butts, large butts, humongous butts, butts having intercourse, butts wearing pants, thongs, panties, etc.
This version is trained on YOLOv8-seg - the instance segmentation model. Basically it is more specific to the part it is targeting instead of a bounding box, it is also trained on 10 additional images than v1 in hopes to help with wet/shiny butts from different angles.
Trained on 310 images, 50 epochs. It will definitely give false positives and sometimes not detect one or two but overall pretty consistent - this is technically the 9th iteration.
The example images are merely showcasing the ability to detect the ass, it is technically prior to the aDetailer being used on it. When using it on a tall image I suggest using a separate width/height to either be square or wider