Spaces:

njgroene

/

age-gender-profilepic

like 1

Runtime error

Hi, could you share the input image used ? It would help everyone get a better understanding of performance. Thanks 🤗

Thanks for sharing, and hope I can offer some explanation! 🤗

I think the result you mention was based of the face reflected in the left of the sunglasses. Due to the low resolution, the best guess output was female (I think at that scale one would probably need a full Gandalf style beard for it to be picked up😅)

The current version of the demo performs input checks, and now rejects this image before classification (for there being multiple faces in the image).

Hope this adds some context! 🙌