Compositional Image Decomposition with Diffusion Models

Jocelin Su, Nan Liu, Joshua B. Tenenbaum, Yilun Du

Primary Area: generative models

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Keywords: Image Decomposition, Compositional Decomposition

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

TL;DR: We present a way to decompose images into sets of composable components.

Abstract: Given an image of a natural scene, we are able to quickly decompose it into a set of components such as objects, lighting, shadows, and foreground. We can then picture how the image would look if we were to recombine certain components with those from other images, for instance producing a scene with a set of objects from our bedroom and animals from a zoo under the lighting conditions of a forest even if we have never seen such a scene in real life before. We present a method to decompose an image into such compositional components. Our approach, Decomp Diffusion, is an unsupervised method which, when given a single image, infers a set of different components in the image, each represented by a diffusion model. We demonstrate how components can capture different factors of the scene, ranging from global scene descriptors (shadows, foreground, facial expression) to local scene descriptors (objects). We further illustrate how inferred factors can be flexibly composed, even with factors inferred from other models, to generate a variety of scenes sharply different than those seen in training time.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 4201