OpenRR-1k: A Scalable Dataset for Real-World Reflection Removal

2025-06-10

AI Summary

Existing reflection removal methods suffer from insufficient generalization due to the scarcity of high-quality, real-world paired data. To address this, we propose OpenRR-1k—the first large-scale, in-the-wild, pixel-aligned reflection removal dataset comprising 1,000 high-fidelity transmission-reflection image pairs. We introduce a novel dual-camera synchronized acquisition and geometric calibration alignment paradigm, uniquely achieving both cost-efficiency, scalability, natural scene diversity, and sub-pixel alignment accuracy. All images are captured across diverse real-world scenes and undergo rigorous quality control. Experiments demonstrate that models trained on OpenRR-1k achieve significantly improved robustness in complex, realistic environments compared to those trained on prior datasets. OpenRR-1k is publicly released and establishes a new benchmark for reflection removal research.

Technology Category

• Computer Vision: Computational Photography, Image & Video Synthesis
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Security and Privacy: Data transparency and provenance
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Reflection removal technology plays a crucial role in photography and computer vision applications. However, existing techniques are hindered by the lack of high-quality in-the-wild datasets. In this paper, we propose a novel paradigm for collecting reflection datasets from a fresh perspective. Our approach is convenient, cost-effective, and scalable, while ensuring that the collected data pairs are of high quality, perfectly aligned, and represent natural and diverse scenarios. Following this paradigm, we collect a Real-world, Diverse, and Pixel-aligned dataset (named OpenRR-1k dataset), which contains 1,000 high-quality transmission-reflection image pairs collected in the wild. Through the analysis of several reflection removal methods and benchmark evaluation experiments on our dataset, we demonstrate its effectiveness in improving robustness in challenging real-world environments. Our dataset is available at https://github.com/caijie0620/OpenRR-1k.

Problem

• Lack of high-quality real-world reflection removal datasets
• Need for scalable diverse pixel-aligned image pairs
• Improving robustness in real-world reflection removal

Innovation

• Novel paradigm for collecting reflection datasets
• Convenient, cost-effective, scalable data collection
• Real-world, diverse, pixel-aligned image pairs