NTIRE 2024 Restore Any Image Model (RAIM) in the Wild Challenge
Jie Liang, Radu Timofte, Qiaosi Yi, Shuaizheng Liu, Lingchen Sun, Rongyuan Wu, Xindong Zhang, Hui Zeng, Lei Zhang
TL;DR
The NTIRE 2024 RAIM in the Wild Challenge tackles restoration of real-world images under unknown, mixed degradations by providing paired and unpaired data streams and a mixed objective/subjective evaluation protocol. The paper details a three-phase workflow, dataset policies, and performance benchmarks, and reports diverse team approaches that combine diffusion priors, GAN-based degradation, and fusion networks to improve perceptual fidelity and generalization. Key contributions include a comprehensive multi-team methodology landscape, phase-wise evaluation results (Phase 2 quantitative and Phase 3 subjective), and practical insights for bridging academic RAIM methods with real-world ISP pipelines. The benchmarks and findings offer a practical foundation for developing robust RAIM models that generalize to diverse, real-world imaging scenarios, with implications for consumer photography and industrial imaging workflows.
Abstract
In this paper, we review the NTIRE 2024 challenge on Restore Any Image Model (RAIM) in the Wild. The RAIM challenge constructed a benchmark for image restoration in the wild, including real-world images with/without reference ground truth in various scenarios from real applications. The participants were required to restore the real-captured images from complex and unknown degradation, where generative perceptual quality and fidelity are desired in the restoration result. The challenge consisted of two tasks. Task one employed real referenced data pairs, where quantitative evaluation is available. Task two used unpaired images, and a comprehensive user study was conducted. The challenge attracted more than 200 registrations, where 39 of them submitted results with more than 400 submissions. Top-ranked methods improved the state-of-the-art restoration performance and obtained unanimous recognition from all 18 judges. The proposed datasets are available at https://drive.google.com/file/d/1DqbxUoiUqkAIkExu3jZAqoElr_nu1IXb/view?usp=sharing and the homepage of this challenge is at https://codalab.lisn.upsaclay.fr/competitions/17632.