Discriminator-Guided Adaptive Diffusion for Source-Free Test-Time Adaptation under Image Corruptions

2026-04-26

AI Summary

This work addresses the performance degradation in source-free unsupervised domain adaptation caused by natural image corruptions—such as blur, weather effects, and digital artifacts—and proposes a test-time, input-level adaptation framework. Leveraging a diffusion model pretrained on the source domain as a generative prior, the method employs a discriminator to dynamically adjust the forward diffusion depth for each test sample. This enables effective removal of corruption-induced artifacts while preserving class-discriminative structures. The denoised image is then reconstructed via reverse diffusion to align with the source domain, allowing inference using a frozen classifier. Experiments across 15 corruption types demonstrate that the approach significantly enhances robustness without sacrificing generalizability, consistently outperforming existing methods—particularly on non-noise corruptions.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

In this work, we study Source-Free Unsupervised Domain Adaptation under corruption-induced domain shifts, where performance degradation is caused by natural image corruptions that go beyond additive noise, including blur, weather effects, and digital artifacts. We propose a diffusion-based, input-level adaptation framework that operates entirely at test time and keeps all source-trained models frozen, explicitly targeting robustness to corrupted target inputs. Our method leverages a source-trained diffusion model as a generative prior and introduces a discriminator-guided adaptive diffusion strategy that dynamically controls the amount of perturbation applied to each test sample. Rather than relying on a fixed diffusion depth, the discriminator determines, on a per-image basis, when sufficient forward diffusion has been applied to suppress corruption-specific artifacts, with each corruption type effectively defining a distinct target domain. This adaptive stopping mechanism applies only the necessary amount of noise to remove domainspecific corruption while preserving class-discriminative structure. The reverse diffusion process then reconstructs a source-aligned image, optionally stabilized through structural guidance, which is classified using a frozen source-trained classifier. We evaluate the proposed approach across a broad spectrum of corruption-induced target domains, covering 15 diverse corruption types, and demonstrate more balanced robustness with competitive or improved performance across non-noise corruptions. Additional analyses reveal how the adaptive diffusion schedule responds to different corruption characteristics, highlighting the practicality, generality, and robustness of the proposed framework. The code is publicly available at https://github.com/fmolivato/dgadiffusion/.

Problem

• Source-Free Unsupervised Domain Adaptation
• Image Corruptions
• Domain Shift
• Test-Time Adaptation
• Robustness

Innovation

• source-free test-time adaptation
• adaptive diffusion
• discriminator-guided
• corruption robustness
• diffusion-based domain adaptation