DiffCut: Catalyzing Zero-Shot Semantic Segmentation with Diffusion Features and Recursive Normalized Cut

TL;DR

This paper introduces DiffCut, a zero-shot image segmentation method using diffusion model features and recursive Normalized Cut, achieving state-of-the-art results without supervision.

Contribution

It leverages diffusion UNet encoder features and recursive Normalized Cut for unsupervised, zero-shot segmentation, outperforming previous methods.

Findings

DiffCut outperforms prior state-of-the-art zero-shot segmentation methods.

Diffusion features effectively encode semantic information for segmentation.

Recursive Normalized Cut improves object granularity and detail in segmentation maps.

Abstract

Foundation models have emerged as powerful tools across various domains including language, vision, and multimodal tasks. While prior works have addressed unsupervised image segmentation, they significantly lag behind supervised models. In this paper, we use a diffusion UNet encoder as a foundation vision encoder and introduce DiffCut, an unsupervised zero-shot segmentation method that solely harnesses the output features from the final self-attention block. Through extensive experimentation, we demonstrate that the utilization of these diffusion features in a graph based segmentation algorithm, significantly outperforms previous state-of-the-art methods on zero-shot segmentation. Specifically, we leverage a recursive Normalized Cut algorithm that softly regulates the granularity of detected objects and produces well-defined segmentation maps that precisely capture intricate image details. Our work highlights the remarkably accurate semantic knowledge embedded within diffusion UNet encoders that could then serve as foundation vision encoders for downstream tasks. Project page at https://diffcut-segmentation.github.io