DarkVRAI: Capture-Condition Conditioning and Burst-Order Selective Scan for Low-light RAW Video Denoising

2025-08-31

AI Summary

To address severe signal degradation in low-light RAW video—caused by high ISO gain and short exposure times under real-time frame-rate constraints—this paper proposes a capture-condition-guided end-to-end denoising framework. The method explicitly incorporates camera metadata (e.g., ISO, exposure time) into motion estimation, frame alignment, and conditional denoising for the first time. It introduces a Burst-Order Selective Scan mechanism to efficiently model long-range temporal dependencies and jointly optimizes multi-frame alignment, feature fusion, and metadata-driven denoising within the RAW domain. Evaluated on the AIM 2025 Low-Light RAW Video Denoising Challenge, our approach ranks first, significantly outperforming state-of-the-art methods on real-world complex scenes and establishing new performance benchmarks.

Technology Category

• Computer Vision: Motion & Tracking
• Search and Optimization: Learning to Search
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Security and Privacy: Data transparency and provenance

Abstract

Low-light RAW video denoising is a fundamentally challenging task due to severe signal degradation caused by high sensor gain and short exposure times, which are inherently limited by video frame rate requirements. To address this, we propose DarkVRAI, a novel framework that achieved first place in the AIM 2025 Low-light RAW Video Denoising Challenge. Our method introduces two primary contributions: (1) a successful application of a conditioning scheme for image denoising, which explicitly leverages capture metadata, to video denoising to guide the alignment and denoising processes, and (2) a Burst-Order Selective Scan (BOSS) mechanism that effectively models long-range temporal dependencies within the noisy video sequence. By synergistically combining these components, DarkVRAI demonstrates state-of-the-art performance on a rigorous and realistic benchmark dataset, setting a new standard for low-light video denoising.

Problem

• Addresses severe signal degradation in low-light RAW videos
• Leverages capture metadata to guide alignment and denoising
• Models long-range temporal dependencies in noisy video sequences

Innovation

• Capture-conditioning scheme using metadata
• Burst-Order Selective Scan mechanism
• Modeling long-range temporal dependencies