Harnessing Diffusion Models for Visual Perception with Meta Prompts

TL;DR

This paper introduces a novel method to adapt pre-trained diffusion models for visual perception tasks by using learnable meta prompts and a recurrent refinement strategy, achieving state-of-the-art results across multiple benchmarks.

Contribution

The paper proposes the use of learnable meta prompts and a recurrent refinement training strategy to effectively repurpose diffusion models for various visual perception tasks.

Findings

Achieves new records in depth estimation on NYU depth V2 and KITTI.

Attains top performance in semantic segmentation on CityScapes.

Performs comparably to state-of-the-art in semantic segmentation on ADE20K and pose estimation on COCO.

Abstract

The issue of generative pretraining for vision models has persisted as a long-standing conundrum. At present, the text-to-image (T2I) diffusion model demonstrates remarkable proficiency in generating high-definition images matching textual inputs, a feat made possible through its pre-training on large-scale image-text pairs. This leads to a natural inquiry: can diffusion models be utilized to tackle visual perception tasks? In this paper, we propose a simple yet effective scheme to harness a diffusion model for visual perception tasks. Our key insight is to introduce learnable embeddings (meta prompts) to the pre-trained diffusion models to extract proper features for perception. The effect of meta prompts are two-fold. First, as a direct replacement of the text embeddings in the T2I models, it can activate task-relevant features during feature extraction. Second, it will be used to re-arrange the extracted features to ensures that the model focuses on the most pertinent features for the task on hand. Additionally, we design a recurrent refinement training strategy that fully leverages the property of diffusion models, thereby yielding stronger visual features. Extensive experiments across various benchmarks validate the effectiveness of our approach. Our approach achieves new performance records in depth estimation tasks on NYU depth V2 and KITTI, and in semantic segmentation task on CityScapes. Concurrently, the proposed method attains results comparable to the current state-of-the-art in semantic segmentation on ADE20K and pose estimation on COCO datasets, further exemplifying its robustness and versatility.