InteractDiffusion: Interaction Control in Text-to-Image Diffusion Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hoe, Jiun Tian, Jiang, Xudong, Chan, Chee Seng, Tan, Yap-Peng, Hu, Weipeng
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917598793302016
author Hoe, Jiun Tian
Jiang, Xudong
Chan, Chee Seng
Tan, Yap-Peng
Hu, Weipeng
author_facet Hoe, Jiun Tian
Jiang, Xudong
Chan, Chee Seng
Tan, Yap-Peng
Hu, Weipeng
contents Large-scale text-to-image (T2I) diffusion models have showcased incredible capabilities in generating coherent images based on textual descriptions, enabling vast applications in content generation. While recent advancements have introduced control over factors such as object localization, posture, and image contours, a crucial gap remains in our ability to control the interactions between objects in the generated content. Well-controlling interactions in generated images could yield meaningful applications, such as creating realistic scenes with interacting characters. In this work, we study the problems of conditioning T2I diffusion models with Human-Object Interaction (HOI) information, consisting of a triplet label (person, action, object) and corresponding bounding boxes. We propose a pluggable interaction control model, called InteractDiffusion that extends existing pre-trained T2I diffusion models to enable them being better conditioned on interactions. Specifically, we tokenize the HOI information and learn their relationships via interaction embeddings. A conditioning self-attention layer is trained to map HOI tokens to visual tokens, thereby conditioning the visual tokens better in existing T2I diffusion models. Our model attains the ability to control the interaction and location on existing T2I diffusion models, which outperforms existing baselines by a large margin in HOI detection score, as well as fidelity in FID and KID. Project page: https://jiuntian.github.io/interactdiffusion.
format Preprint
id arxiv_https___arxiv_org_abs_2312_05849
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle InteractDiffusion: Interaction Control in Text-to-Image Diffusion Models
Hoe, Jiun Tian
Jiang, Xudong
Chan, Chee Seng
Tan, Yap-Peng
Hu, Weipeng
Computer Vision and Pattern Recognition
Graphics
Multimedia
Large-scale text-to-image (T2I) diffusion models have showcased incredible capabilities in generating coherent images based on textual descriptions, enabling vast applications in content generation. While recent advancements have introduced control over factors such as object localization, posture, and image contours, a crucial gap remains in our ability to control the interactions between objects in the generated content. Well-controlling interactions in generated images could yield meaningful applications, such as creating realistic scenes with interacting characters. In this work, we study the problems of conditioning T2I diffusion models with Human-Object Interaction (HOI) information, consisting of a triplet label (person, action, object) and corresponding bounding boxes. We propose a pluggable interaction control model, called InteractDiffusion that extends existing pre-trained T2I diffusion models to enable them being better conditioned on interactions. Specifically, we tokenize the HOI information and learn their relationships via interaction embeddings. A conditioning self-attention layer is trained to map HOI tokens to visual tokens, thereby conditioning the visual tokens better in existing T2I diffusion models. Our model attains the ability to control the interaction and location on existing T2I diffusion models, which outperforms existing baselines by a large margin in HOI detection score, as well as fidelity in FID and KID. Project page: https://jiuntian.github.io/interactdiffusion.
title InteractDiffusion: Interaction Control in Text-to-Image Diffusion Models
topic Computer Vision and Pattern Recognition
Graphics
Multimedia
url https://arxiv.org/abs/2312.05849