showlab / DatasetDM

[NeurIPS2023] DatasetDM:Synthesizing Data with Perception Annotations Using Diffusion Models
https://weijiawu.github.io/DatasetDM_page/
299 stars 12 forks source link

add noise problem #4

Closed fubenzun closed 1 year ago

fubenzun commented 1 year ago

thanks for your work. in trainng stage,why add 1 step noise to train P-decoder

weijiawu commented 1 year ago

thanks for your attention. This is the fundamental principle of the diffusion process, which involves adding noise first and then employing a U-Net to predict the noise.

fubenzun commented 1 year ago

Comparing the method of generating images and segmentation maps using a diffusion model with the method of generating images using a diffusion model and then passing the images through a Segment Anything Model, which one is more accurate