FOCUS: Frequency-Optimized Conditioning of DiffUSion Models for mitigating catastrophic forgetting during Test-Time Adaptation

TL;DR

FOCUS introduces a frequency-based conditioning method for diffusion models that enhances test-time adaptation by preserving task-relevant information and mitigating catastrophic forgetting across diverse corruptions.

Contribution

The paper proposes FOCUS, a novel frequency-based conditioning approach with a lightweight network and data augmentation, improving domain adaptation and reducing forgetting in diffusion models.

Findings

Achieves state-of-the-art performance in semantic segmentation and depth estimation across 15 corruptions.

Enhances existing adaptation methods by providing pseudo labels from denoised images.

Reduces computational costs with a lightweight frequency prediction network.

Abstract

Test-time adaptation enables models to adapt to evolving domains. However, balancing the tradeoff between preserving knowledge and adapting to domain shifts remains challenging for model adaptation methods, since adapting to domain shifts can induce forgetting of task-relevant knowledge. To address this problem, we propose FOCUS, a novel frequency-based conditioning approach within a diffusion-driven input-adaptation framework. Utilising learned, spatially adaptive frequency priors, our approach conditions the reverse steps during diffusion-driven denoising to preserve task-relevant semantic information for dense prediction. FOCUS leverages a trained, lightweight, Y-shaped Frequency Prediction Network (Y-FPN) that disentangles high and low frequency information from noisy images. This minimizes the computational costs involved in implementing our approach in a diffusion-driven framework. We train Y-FPN with FrequencyMix, a novel data augmentation method that perturbs the images across diverse frequency bands, which improves the robustness of our approach to diverse corruptions. We demonstrate the effectiveness of FOCUS for semantic segmentation and monocular depth estimation across 15 corruption types and three datasets, achieving state-of-the-art averaged performance. In addition to improving standalone performance, FOCUS complements existing model adaptation methods since we can derive pseudo labels from FOCUS-denoised images for additional supervision. Even under limited, intermittent supervision with the pseudo labels derived from the FOCUS denoised images, we show that FOCUS mitigates catastrophic forgetting for recent model adaptation methods.