Which model ckpt of SD used in the paper for finetune, and how to load the model ckpt of SD in the main.py

IDEA-Research / HumanSD

[ICCV 2023] The official implementation of paper "HumanSD: A Native Skeleton-Guided Diffusion Model for Human Image Generation"

Apache License 2.0

271 stars 18 forks source link

Which model ckpt of SD used in the paper for finetune, and how to load the model ckpt of SD in the main.py #13

Closed ucasyjz closed 1 year ago

ucasyjz commented 1 year ago

Can you give me some guidance, thank you

juxuan27 commented 1 year ago

Hi, @yang326922943 ! Thank you for your focus. We directly use stable diffusion v2.1's checkpoint and you can directly load the checkpoint using the given code (with the provided yaml file).

ucasyjz commented 1 year ago

Thanks, I have another question, how the Quantitative Results are obtained, on which dataset and with which python file

juxuan27 commented 1 year ago

Please refer to the supplementary file for detail calculation of Quantitative Results. We mainly use torchmetrics and COCOAPI to calculate the metrics. The code will be release as soon as possible.

juxuan27 commented 1 year ago

Updated at https://github.com/IDEA-Research/HumanSD/tree/main/utils/metrics