CleanDIFT: Diffusion Features without Noise

TL;DR

This paper introduces a lightweight unsupervised fine-tuning approach for diffusion models that produces high-quality, noise-free semantic features, significantly improving downstream task performance without the need for noise addition or ensembling.

Contribution

It presents a novel fine-tuning method enabling diffusion models to generate effective noise-free features, surpassing previous noisy feature extraction techniques.

Findings

Outperforms previous diffusion features in various tasks

Requires less computational cost than ensemble methods

Produces high-quality, noise-free semantic features

Abstract

Internal features from large-scale pre-trained diffusion models have recently been established as powerful semantic descriptors for a wide range of downstream tasks. Works that use these features generally need to add noise to images before passing them through the model to obtain the semantic features, as the models do not offer the most useful features when given images with little to no noise. We show that this noise has a critical impact on the usefulness of these features that cannot be remedied by ensembling with different random noises. We address this issue by introducing a lightweight, unsupervised fine-tuning method that enables diffusion backbones to provide high-quality, noise-free semantic features. We show that these features readily outperform previous diffusion features by a wide margin in a wide variety of extraction setups and downstream tasks, offering better performance than even ensemble-based methods at a fraction of the cost.