Where's Waldo: Diffusion Features for Personalized Segmentation and Retrieval

TL;DR

This paper introduces PDM, a novel method leveraging diffusion model features for personalized image retrieval and segmentation, outperforming supervised methods without additional training.

Contribution

The paper proposes PDM, a new approach using diffusion model features for personalized tasks, addressing limitations of current models and introducing new benchmarks.

Findings

PDM outperforms supervised methods on key benchmarks.

Diffusion features enable effective personalization without training.

Identifies dataset shortcomings and proposes new benchmarks.

Abstract

Personalized retrieval and segmentation aim to locate specific instances within a dataset based on an input image and a short description of the reference instance. While supervised methods are effective, they require extensive labeled data for training. Recently, self-supervised foundation models have been introduced to these tasks showing comparable results to supervised methods. However, a significant flaw in these models is evident: they struggle to locate a desired instance when other instances within the same class are presented. In this paper, we explore text-to-image diffusion models for these tasks. Specifically, we propose a novel approach called PDM for Personalized Features Diffusion Matching, that leverages intermediate features of pre-trained text-to-image models for personalization tasks without any additional training. PDM demonstrates superior performance on popular retrieval and segmentation benchmarks, outperforming even supervised methods. We also highlight notable shortcomings in current instance and segmentation datasets and propose new benchmarks for these tasks.