search

little-misfit / BooW-VTON

Offical code repository of “BooW-VTON: Boosting In-the-Wild Virtual Try-On via Mask-Free Pseudo Data Training”
MIT License
67 stars 0 forks source link

attention map display #11

Closed Wangxuefeng92 closed 1 month ago

Wangxuefeng92 commented 1 month ago

hi, sir. could you share something about 'attention map display' like Figure4 in your paper ? thank you for your great work on BooW-VTON!

little-misfit commented 1 month ago

You can refer to FastComposer: Tuning-Free Multi-Subject Image Generation with Localized Attention 's code about how to extract the attention map. For virtual try-on task, multiple attention maps need to be averaged into one.