About: Foot | Detection for Adetailer is a custom trained YOLOv8 model meant to be used with ADetailer or any other bbox/detection compatible plugin for Automatic1111 or ComfyUI.
Detection: This model can detect and segment human foot and feet, both barefoot and with clothes on.
Segm model you can find in my Telegram channel
Recommended BBOX Treshold ≥ 0.5
Description
FAQ
Comments (11)
Seriously underrated.
For ComfyUI, where do I put this??
I tried adding it to the ultralytics/bbox folder, but it's not showing up in my comfyUI.
it's BBOX version. Put it in ComfyUI_windows_portable\ComfyUI\models\ultralytics\bbox. SEGM version in my Tg
@Spoony my bad, i was adding it to the wrong folder.
This deserves a report for telegram link, frl, un-downloadable without telegram account and telegram wants your phone number, which you should not do under any circumstances as phone number has your data linked to it and telegram is a nest for scamers.
Multibillion dollar companies have already sold all your personal data many times over, yet you’re afraid to register in one of the largest messengers. With that logic, you might as well never go outside either, since there are scammers everywhere
@degurshaft I don't think that excuses anything, I should not have to go offsite to a service I don't want to make an account for, for the full download. Also just because several companies have sold my data does not mean i want it sold many times more.
@yakasuki As for the model author, yeah, that doesn’t excuse him at all. It is super tacky to poach users like that. But honestly, being suspicious of one of the world’s biggest messengers is pretty weird
Honestly, the model doesn’t handle its task very well. It seems the dataset is either too small or not diverse enough. A model made by another person for hand detection actually detects feets better than this one
Where is the Segm download?
It often turns arms, knees, and whole people into legs. I've already set the trigger threshold to 0.65, but the model looks at the arm and says, "Yes, there's a 0.67 probability that it's a leg!" However, two feet in another part of the image were overlooked.
I don't recommend it.