search

facebookresearch / sapiens

High-resolution models for human tasks.
https://about.meta.com/realitylabs/codecavatars/sapiens/
Other
4.34k stars 234 forks source link

Question about input resolution #153

Closed yk907 closed 2 hours ago

yk907 commented 2 hours ago

Thank you for your excellent work!

I am using your lite image encoder for my downstream task inference, and the image resolution in my dataset is only 448 224. When I am using the Lite version sapiens for inference, I report an error that the dimensions do not match. Do I have to use a resolution of 1024 1024 for image encoding? Or do I need to install the full version of Sapiens or any suggestions? thanks.

rawalkhirodkar commented 2 hours ago

@yk907 please do not update the input size, it corresponds to the model size. Any size image will be resized to 1024 x 768 to support inference with Sapiens.