QualiTeacher: Quality-Conditioned Pseudo-Labeling for Real-World Image Restoration

2026-03-09

AI Summary

Real-world image restoration often relies on pseudo-labels due to the absence of ground-truth clean images, yet low-quality pseudo-labels can introduce artifacts and hinder generalization. To address this, this work proposes QualiTeacher, a novel framework that leverages non-reference image quality assessment (NR-IQA) models to estimate the quality of pseudo-labels and uses these estimates as conditional supervision signals to guide the student network in learning a quality-aware restoration manifold. The method integrates multi-dimensional NR-IQA metrics, diverse augmentation strategies, a preference-based monotonic quality ranking loss inspired by DPO, and a cropping consistency loss, effectively transforming pseudo-label quality from a source of noise into informative supervision. Experiments demonstrate that QualiTeacher significantly improves performance on standard real-world restoration benchmarks and functions as a plug-and-play module that enhances existing pseudo-labeling approaches—even surpassing the teacher model itself.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Diffusion Models for Vision
• Knowledge Representation and Reasoning: Qualitative Reasoning

Application Category

• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Real-world image restoration (RWIR) is a highly challenging task due to the absence of clean ground-truth images. Many recent methods resort to pseudo-label (PL) supervision, often within a Mean-Teacher (MT) framework. However, these methods face a critical paradox: unconditionally trusting the often imperfect, low-quality PLs forces the student model to learn undesirable artifacts, while discarding them severely limits data diversity and impairs model generalization. In this paper, we propose QualiTeacher, a novel framework that transforms pseudo-label quality from a noisy liability into a conditional supervisory signal. Instead of filtering, QualiTeacher explicitly conditions the student model on the quality of the PLs, estimated by an ensemble of complementary non-reference image quality assessment (NR-IQA) models spanning low-level distortion and semantic-level assessment. This strategy teaches the student network to learn a quality-graded restoration manifold, enabling it to understand what constitutes different quality levels. Consequently, it can not only avoid mimicking artifacts from low-quality labels but also extrapolate to generate results of higher quality than the teacher itself. To ensure the robustness and accuracy of this quality-driven learning, we further enhance the process with a multi-augmentation scheme to diversify the PL quality spectrum, a score-based preference optimization strategy inspired by Direct Preference Optimization (DPO) to enforce a monotonically ordered quality separation, and a cropped consistency loss to prevent adversarial over-optimization (reward hacking) of the IQA models. Experiments on standard RWIR benchmarks demonstrate that QualiTeacher can serve as a plug-and-play strategy to improve the quality of the existing pseudo-labeling framework, establishing a new paradigm for learning from imperfect supervision. Code will be released.

Problem

• Real-world image restoration
• Pseudo-labeling
• Image quality assessment
• Imperfect supervision
• Mean-Teacher framework

Innovation

• Quality-Conditioned Pseudo-Labeling
• Mean-Teacher Framework
• Non-Reference Image Quality Assessment
• Preference Optimization
• Real-World Image Restoration