Skip to yearly menu bar Skip to main content


Grounded Text-to-Image Synthesis with Attention Refocusing

Quynh Phung · Songwei Ge · Jia-Bin Huang

Arch 4A-E Poster #304
[ ]
Wed 19 Jun 5 p.m. PDT — 6:30 p.m. PDT


Driven by the scalable diffusion models trained on large-scale datasets, text-to-image synthesis methods have shown compelling results. However, these models still fail to precisely follow the text prompt involving multiple objects, attributes, or spatial compositions.In this paper, we reveal the potential causes of the diffusion model's cross-attention and self-attention layers. We propose two novel losses to refocus attention maps according to a given spatial layout during sampling. Creating the layouts manually requires additional effort and can be tedious. Therefore, we explore using large language models (LLM) to produce these layouts for our method. We conduct extensive experiments on the DrawBench, HRS, and TIFA benchmarks to evaluate our proposed method. We show that our proposed attention refocusing effectively improves the controllability of existing approaches.

Live content is unavailable. Log in and register to view live content