search

weijiawu / DiffuMask

[ICCV2023] DiffuMask: Synthesizing Images with Pixel-level Annotations for Semantic Segmentation Using Diffusion Models
150 stars 11 forks source link

code? #1

Open jiangyuanlin01 opened 1 year ago

jiangyuanlin01 commented 1 year ago

can you give me your code?

weijiawu commented 1 year ago

Sorry, the paper is under review, we have no plan to release the code before the result.

weijiawu commented 1 year ago

But the method is quite simple, you can implement it by yourself.

WANGSSSSSSS commented 1 year ago

@weijiawu hi, The original Affinity net swallows a rgb image, generates a affinity matrix, and use this matrix to refine the cam results to obtain a binary mask for each class, Right? But the paper present a weird view: image confused 😭 about this input ⬆️

weijiawu commented 1 year ago

@weijiawu hi, The original Affinity net swallows a rgb image, generates a affinity matrix, and use this matrix to refine the cam results to obtain a binary mask for each class, Right? But the paper present a weird view: image confused 😭 about this input ⬆️

Sure, similar to the original code, the image, and binary mask are all needed for the affinityNet. We follow the code (https://github.com/jiwoon-ahn/psa).

Help that can help you, 😊