Closed Usernamezhx closed 8 months ago
Hi, thank you for your interest in our projects!
LayoutDM+ inserts the cross-attention layer (typical one in Transformer decoder) into the building block to attend flattened image feature maps.
thanks for your reply.
Thank you for your contribution to the layout work. I read the paper "Retrieval-Augmented Layout Transformer". it point that:
LayoutDM† [19] is a discrete state-space diffusion model that can handle many constrained generation tasks. Since the model is originally designed for content-agnostic layout generation, we extend the model to accept an input image.
Can you provide some ideas about the layoutDM+. thanks very much.