NTIRE 2025 Challenge on Real-World Face Restoration: Methods and Results

2025-04-20

AI Summary

This work addresses facial image restoration under real-world conditions, aiming to simultaneously preserve visual naturalness and identity fidelity to enhance perceptual quality and realism. We propose an end-to-end framework integrating multi-scale reconstruction, GAN-based fine-detail synthesis, and identity-aware loss optimization. To rigorously evaluate performance, we introduce— for the first time—a joint assessment protocol combining weighted image quality assessment (IQA) and AdaFace-based identity verification, enabling practical performance breakthroughs without computational or dataset-scale constraints. The method significantly improves restoration fidelity and cross-domain generalization capability. Validated in a large-scale international competition, it attracted 141 participating teams, with 10 advancing to final rankings—demonstrating both effectiveness and state-of-the-art advancement in facial restoration.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Security and Privacy: Large-scale security measurements

Abstract

This paper provides a review of the NTIRE 2025 challenge on real-world face restoration, highlighting the proposed solutions and the resulting outcomes. The challenge focuses on generating natural, realistic outputs while maintaining identity consistency. Its goal is to advance state-of-the-art solutions for perceptual quality and realism, without imposing constraints on computational resources or training data. The track of the challenge evaluates performance using a weighted image quality assessment (IQA) score and employs the AdaFace model as an identity checker. The competition attracted 141 registrants, with 13 teams submitting valid models, and ultimately, 10 teams achieved a valid score in the final ranking. This collaborative effort advances the performance of real-world face restoration while offering an in-depth overview of the latest trends in the field.

Problem

• Advance real-world face restoration techniques
• Ensure identity consistency in restored faces
• Improve perceptual quality and realism outputs

Innovation

• Uses AdaFace model for identity verification
• Employs weighted IQA score for evaluation
• No constraints on computational resources